diff --git a/MANIFEST.in b/MANIFEST.in
index d5a30fc..08e1547 100644
--- a/MANIFEST.in
+++ b/MANIFEST.in
@@ -1,4 +1,5 @@
 include README.md LICENSE
 include bayesclass/FeatureSelect.h
 include bayesclass/Node.h
-include bayesclass/Network.h
\ No newline at end of file
+include bayesclass/Network.h
+include bayesclass/Metrics.hpp
\ No newline at end of file
diff --git a/bayesclass/BayesNetwork.cpp b/bayesclass/BayesNetwork.cpp
index fc63c38..5c5d4ec 100644
--- a/bayesclass/BayesNetwork.cpp
+++ b/bayesclass/BayesNetwork.cpp
@@ -769,6 +769,7 @@ static CYTHON_INLINE float __PYX_NAN() {
 #include <string.h>
 #include <string>
 #include "Network.h"
+#include "Metrics.hpp"
 #ifdef _OPENMP
 #include <omp.h>
 #endif /* _OPENMP */
@@ -984,8 +985,9 @@ static const char *__pyx_f[] = {
 
 /*--- Type declarations ---*/
 struct __pyx_obj_10bayesclass_8BayesNet_BayesNetwork;
+struct __pyx_obj_10bayesclass_8BayesNet_CMetrics;
 
-/* "bayesclass/BayesNetwork.pyx":22
+/* "bayesclass/BayesNetwork.pyx":21
  *         string version()
  * 
  * cdef class BayesNetwork:             # <<<<<<<<<<<<<<
@@ -998,6 +1000,19 @@ struct __pyx_obj_10bayesclass_8BayesNet_BayesNetwork {
 };
 
 
+/* "bayesclass/BayesNetwork.pyx":63
+ *         vector[float] test()
+ * 
+ * cdef class CMetrics:             # <<<<<<<<<<<<<<
+ *     cdef Metrics *thisptr
+ *     def __cinit__(self, X, y, features, className, classStates):
+ */
+struct __pyx_obj_10bayesclass_8BayesNet_CMetrics {
+  PyObject_HEAD
+  bayesnet::Metrics *thisptr;
+};
+
+
 /* --- Runtime support code (head) --- */
 /* Refnanny.proto */
 #ifndef CYTHON_REFNANNY
@@ -1417,6 +1432,7 @@ static int __Pyx_InitStrings(__Pyx_StringTabEntry *t);
 
 /* Module declarations from 'bayesclass.BayesNet' */
 static PyTypeObject *__pyx_ptype_10bayesclass_8BayesNet_BayesNetwork = 0;
+static PyTypeObject *__pyx_ptype_10bayesclass_8BayesNet_CMetrics = 0;
 static std::vector<int>  __pyx_convert_vector_from_py_int(PyObject *); /*proto*/
 static std::vector<std::vector<int> >  __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(PyObject *); /*proto*/
 static std::string __pyx_convert_string_from_py_std__in_string(PyObject *); /*proto*/
@@ -1429,6 +1445,7 @@ static CYTHON_INLINE PyObject *__pyx_convert_PyUnicode_string_to_py_std__in_stri
 static CYTHON_INLINE PyObject *__pyx_convert_PyStr_string_to_py_std__in_string(std::string const &); /*proto*/
 static CYTHON_INLINE PyObject *__pyx_convert_PyBytes_string_to_py_std__in_string(std::string const &); /*proto*/
 static CYTHON_INLINE PyObject *__pyx_convert_PyByteArray_string_to_py_std__in_string(std::string const &); /*proto*/
+static PyObject *__pyx_convert_vector_to_py_float(const std::vector<float>  &); /*proto*/
 #define __Pyx_MODULE_NAME "bayesclass.BayesNet"
 extern int __pyx_module_is_main_bayesclass__BayesNet;
 int __pyx_module_is_main_bayesclass__BayesNet = 0;
@@ -1446,16 +1463,20 @@ static const char __pyx_k_encode[] = "encode";
 static const char __pyx_k_name_2[] = "__name__";
 static const char __pyx_k_source[] = "source";
 static const char __pyx_k_states[] = "states";
+static const char __pyx_k_CMetrics[] = "CMetrics";
 static const char __pyx_k_features[] = "features";
 static const char __pyx_k_className[] = "className";
 static const char __pyx_k_maxThreads[] = "maxThreads";
+static const char __pyx_k_classStates[] = "classStates";
 static const char __pyx_k_destination[] = "destination";
 static const char __pyx_k_BayesNetwork[] = "BayesNetwork";
 static const char __pyx_k_laplaceSmooth[] = "laplaceSmooth";
 static const char __pyx_k_cline_in_traceback[] = "cline_in_traceback";
 static PyObject *__pyx_n_s_BayesNetwork;
+static PyObject *__pyx_n_s_CMetrics;
 static PyObject *__pyx_n_s_X;
 static PyObject *__pyx_n_s_className;
+static PyObject *__pyx_n_s_classStates;
 static PyObject *__pyx_n_s_cline_in_traceback;
 static PyObject *__pyx_n_s_destination;
 static PyObject *__pyx_n_s_encode;
@@ -1484,13 +1505,19 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_18getStates(stru
 static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_20getClassName(struct __pyx_obj_10bayesclass_8BayesNet_BayesNetwork *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_22getClassNumStates(struct __pyx_obj_10bayesclass_8BayesNet_BayesNetwork *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_24__reduce__(CYTHON_UNUSED struct __pyx_obj_10bayesclass_8BayesNet_BayesNetwork *__pyx_v_self); /* proto */
+static int __pyx_pf_10bayesclass_8BayesNet_8CMetrics___cinit__(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self, PyObject *__pyx_v_X, PyObject *__pyx_v_y, PyObject *__pyx_v_features, PyObject *__pyx_v_className, PyObject *__pyx_v_classStates); /* proto */
+static void __pyx_pf_10bayesclass_8BayesNet_8CMetrics_2__dealloc__(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self); /* proto */
+static PyObject *__pyx_pf_10bayesclass_8BayesNet_8CMetrics_4conditionalEdgeWeights(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self); /* proto */
+static PyObject *__pyx_pf_10bayesclass_8BayesNet_8CMetrics_6test(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self); /* proto */
+static PyObject *__pyx_pf_10bayesclass_8BayesNet_8CMetrics_8__reduce__(CYTHON_UNUSED struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self); /* proto */
 static PyObject *__pyx_tp_new_10bayesclass_8BayesNet_BayesNetwork(PyTypeObject *t, PyObject *a, PyObject *k); /*proto*/
+static PyObject *__pyx_tp_new_10bayesclass_8BayesNet_CMetrics(PyTypeObject *t, PyObject *a, PyObject *k); /*proto*/
 static PyObject *__pyx_float_0_8;
 static PyObject *__pyx_float_1_0;
 static PyObject *__pyx_slice_;
 /* Late includes */
 
-/* "bayesclass/BayesNetwork.pyx":24
+/* "bayesclass/BayesNetwork.pyx":23
  * cdef class BayesNetwork:
  *     cdef Network *thisptr
  *     def __cinit__(self, maxThreads=0.8, laplaceSmooth=1.0):             # <<<<<<<<<<<<<<
@@ -1540,7 +1567,7 @@ static int __pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_1__cinit__(PyObject *_
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "__cinit__") < 0)) __PYX_ERR(0, 24, __pyx_L3_error)
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "__cinit__") < 0)) __PYX_ERR(0, 23, __pyx_L3_error)
       }
     } else {
       switch (PyTuple_GET_SIZE(__pyx_args)) {
@@ -1557,7 +1584,7 @@ static int __pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_1__cinit__(PyObject *_
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("__cinit__", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 24, __pyx_L3_error)
+  __Pyx_RaiseArgtupleInvalid("__cinit__", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 23, __pyx_L3_error)
   __pyx_L3_error:;
   __Pyx_AddTraceback("bayesclass.BayesNet.BayesNetwork.__cinit__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
@@ -1581,24 +1608,24 @@ static int __pyx_pf_10bayesclass_8BayesNet_12BayesNetwork___cinit__(struct __pyx
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__cinit__", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":25
+  /* "bayesclass/BayesNetwork.pyx":24
  *     cdef Network *thisptr
  *     def __cinit__(self, maxThreads=0.8, laplaceSmooth=1.0):
  *         self.thisptr = new Network(maxThreads, laplaceSmooth)             # <<<<<<<<<<<<<<
  *     def __dealloc__(self):
  *         del self.thisptr
  */
-  __pyx_t_1 = __pyx_PyFloat_AsFloat(__pyx_v_maxThreads); if (unlikely((__pyx_t_1 == (float)-1) && PyErr_Occurred())) __PYX_ERR(0, 25, __pyx_L1_error)
-  __pyx_t_2 = __pyx_PyFloat_AsFloat(__pyx_v_laplaceSmooth); if (unlikely((__pyx_t_2 == (float)-1) && PyErr_Occurred())) __PYX_ERR(0, 25, __pyx_L1_error)
+  __pyx_t_1 = __pyx_PyFloat_AsFloat(__pyx_v_maxThreads); if (unlikely((__pyx_t_1 == (float)-1) && PyErr_Occurred())) __PYX_ERR(0, 24, __pyx_L1_error)
+  __pyx_t_2 = __pyx_PyFloat_AsFloat(__pyx_v_laplaceSmooth); if (unlikely((__pyx_t_2 == (float)-1) && PyErr_Occurred())) __PYX_ERR(0, 24, __pyx_L1_error)
   try {
     __pyx_t_3 = new bayesnet::Network(__pyx_t_1, __pyx_t_2);
   } catch(...) {
     __Pyx_CppExn2PyErr();
-    __PYX_ERR(0, 25, __pyx_L1_error)
+    __PYX_ERR(0, 24, __pyx_L1_error)
   }
   __pyx_v_self->thisptr = __pyx_t_3;
 
-  /* "bayesclass/BayesNetwork.pyx":24
+  /* "bayesclass/BayesNetwork.pyx":23
  * cdef class BayesNetwork:
  *     cdef Network *thisptr
  *     def __cinit__(self, maxThreads=0.8, laplaceSmooth=1.0):             # <<<<<<<<<<<<<<
@@ -1617,7 +1644,7 @@ static int __pyx_pf_10bayesclass_8BayesNet_12BayesNetwork___cinit__(struct __pyx
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":26
+/* "bayesclass/BayesNetwork.pyx":25
  *     def __cinit__(self, maxThreads=0.8, laplaceSmooth=1.0):
  *         self.thisptr = new Network(maxThreads, laplaceSmooth)
  *     def __dealloc__(self):             # <<<<<<<<<<<<<<
@@ -1640,7 +1667,7 @@ static void __pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_2__dealloc__(struct _
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("__dealloc__", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":27
+  /* "bayesclass/BayesNetwork.pyx":26
  *         self.thisptr = new Network(maxThreads, laplaceSmooth)
  *     def __dealloc__(self):
  *         del self.thisptr             # <<<<<<<<<<<<<<
@@ -1649,7 +1676,7 @@ static void __pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_2__dealloc__(struct _
  */
   delete __pyx_v_self->thisptr;
 
-  /* "bayesclass/BayesNetwork.pyx":26
+  /* "bayesclass/BayesNetwork.pyx":25
  *     def __cinit__(self, maxThreads=0.8, laplaceSmooth=1.0):
  *         self.thisptr = new Network(maxThreads, laplaceSmooth)
  *     def __dealloc__(self):             # <<<<<<<<<<<<<<
@@ -1661,7 +1688,7 @@ static void __pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_2__dealloc__(struct _
   __Pyx_RefNannyFinishContext();
 }
 
-/* "bayesclass/BayesNetwork.pyx":28
+/* "bayesclass/BayesNetwork.pyx":27
  *     def __dealloc__(self):
  *         del self.thisptr
  *     def fit(self, X, y, features, className):             # <<<<<<<<<<<<<<
@@ -1709,23 +1736,23 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_5fit(PyObject *_
         case  1:
         if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_y)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, 1); __PYX_ERR(0, 28, __pyx_L3_error)
+          __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, 1); __PYX_ERR(0, 27, __pyx_L3_error)
         }
         CYTHON_FALLTHROUGH;
         case  2:
         if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_features)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, 2); __PYX_ERR(0, 28, __pyx_L3_error)
+          __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, 2); __PYX_ERR(0, 27, __pyx_L3_error)
         }
         CYTHON_FALLTHROUGH;
         case  3:
         if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_className)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, 3); __PYX_ERR(0, 28, __pyx_L3_error)
+          __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, 3); __PYX_ERR(0, 27, __pyx_L3_error)
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "fit") < 0)) __PYX_ERR(0, 28, __pyx_L3_error)
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "fit") < 0)) __PYX_ERR(0, 27, __pyx_L3_error)
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 4) {
       goto __pyx_L5_argtuple_error;
@@ -1742,7 +1769,7 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_5fit(PyObject *_
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 28, __pyx_L3_error)
+  __Pyx_RaiseArgtupleInvalid("fit", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 27, __pyx_L3_error)
   __pyx_L3_error:;
   __Pyx_AddTraceback("bayesclass.BayesNet.BayesNetwork.fit", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
@@ -1778,7 +1805,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("fit", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":29
+  /* "bayesclass/BayesNetwork.pyx":28
  *         del self.thisptr
  *     def fit(self, X, y, features, className):
  *         X_ = [X[:, i] for i in range(X.shape[1])]             # <<<<<<<<<<<<<<
@@ -1786,23 +1813,23 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
  *         self.thisptr.fit(X_, y, features_bytes, className.encode())
  */
   { /* enter inner scope */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 29, __pyx_L5_error)
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 28, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 29, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 28, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 29, __pyx_L5_error)
+    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 28, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 29, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 28, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     if (likely(PyList_CheckExact(__pyx_t_2)) || PyTuple_CheckExact(__pyx_t_2)) {
       __pyx_t_3 = __pyx_t_2; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
       __pyx_t_5 = NULL;
     } else {
-      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 29, __pyx_L5_error)
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 28, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_3);
-      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 29, __pyx_L5_error)
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 28, __pyx_L5_error)
     }
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     for (;;) {
@@ -1810,17 +1837,17 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
         if (likely(PyList_CheckExact(__pyx_t_3))) {
           if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 29, __pyx_L5_error)
+          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 28, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 29, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 28, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         } else {
           if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 29, __pyx_L5_error)
+          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 28, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 29, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 28, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         }
@@ -1830,7 +1857,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
           PyObject* exc_type = PyErr_Occurred();
           if (exc_type) {
             if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
-            else __PYX_ERR(0, 29, __pyx_L5_error)
+            else __PYX_ERR(0, 28, __pyx_L5_error)
           }
           break;
         }
@@ -1838,7 +1865,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
       }
       __Pyx_XDECREF_SET(__pyx_7genexpr__pyx_v_i, __pyx_t_2);
       __pyx_t_2 = 0;
-      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 29, __pyx_L5_error)
+      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 28, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_INCREF(__pyx_slice_);
       __Pyx_GIVEREF(__pyx_slice_);
@@ -1846,10 +1873,10 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
       __Pyx_INCREF(__pyx_7genexpr__pyx_v_i);
       __Pyx_GIVEREF(__pyx_7genexpr__pyx_v_i);
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_7genexpr__pyx_v_i);
-      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 29, __pyx_L5_error)
+      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 28, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_6);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 29, __pyx_L5_error)
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 28, __pyx_L5_error)
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     }
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -1863,7 +1890,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
   __pyx_v_X_ = ((PyObject*)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "bayesclass/BayesNetwork.pyx":30
+  /* "bayesclass/BayesNetwork.pyx":29
  *     def fit(self, X, y, features, className):
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         features_bytes = [x.encode() for x in features]             # <<<<<<<<<<<<<<
@@ -1871,32 +1898,32 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
  *         return self
  */
   { /* enter inner scope */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 30, __pyx_L11_error)
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 29, __pyx_L11_error)
     __Pyx_GOTREF(__pyx_t_1);
     if (likely(PyList_CheckExact(__pyx_v_features)) || PyTuple_CheckExact(__pyx_v_features)) {
       __pyx_t_3 = __pyx_v_features; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
       __pyx_t_5 = NULL;
     } else {
-      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_v_features); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 30, __pyx_L11_error)
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_v_features); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 29, __pyx_L11_error)
       __Pyx_GOTREF(__pyx_t_3);
-      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 30, __pyx_L11_error)
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 29, __pyx_L11_error)
     }
     for (;;) {
       if (likely(!__pyx_t_5)) {
         if (likely(PyList_CheckExact(__pyx_t_3))) {
           if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_6 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_6); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 30, __pyx_L11_error)
+          __pyx_t_6 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_6); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 29, __pyx_L11_error)
           #else
-          __pyx_t_6 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 30, __pyx_L11_error)
+          __pyx_t_6 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 29, __pyx_L11_error)
           __Pyx_GOTREF(__pyx_t_6);
           #endif
         } else {
           if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_6 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_6); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 30, __pyx_L11_error)
+          __pyx_t_6 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_6); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 29, __pyx_L11_error)
           #else
-          __pyx_t_6 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 30, __pyx_L11_error)
+          __pyx_t_6 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 29, __pyx_L11_error)
           __Pyx_GOTREF(__pyx_t_6);
           #endif
         }
@@ -1906,7 +1933,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
           PyObject* exc_type = PyErr_Occurred();
           if (exc_type) {
             if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
-            else __PYX_ERR(0, 30, __pyx_L11_error)
+            else __PYX_ERR(0, 29, __pyx_L11_error)
           }
           break;
         }
@@ -1914,7 +1941,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
       }
       __Pyx_XDECREF_SET(__pyx_8genexpr1__pyx_v_x, __pyx_t_6);
       __pyx_t_6 = 0;
-      __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_8genexpr1__pyx_v_x, __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 30, __pyx_L11_error)
+      __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_8genexpr1__pyx_v_x, __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 29, __pyx_L11_error)
       __Pyx_GOTREF(__pyx_t_2);
       __pyx_t_7 = NULL;
       if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_2))) {
@@ -1928,10 +1955,10 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
       }
       __pyx_t_6 = (__pyx_t_7) ? __Pyx_PyObject_CallOneArg(__pyx_t_2, __pyx_t_7) : __Pyx_PyObject_CallNoArg(__pyx_t_2);
       __Pyx_XDECREF(__pyx_t_7); __pyx_t_7 = 0;
-      if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 30, __pyx_L11_error)
+      if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 29, __pyx_L11_error)
       __Pyx_GOTREF(__pyx_t_6);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 30, __pyx_L11_error)
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 29, __pyx_L11_error)
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     }
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -1945,17 +1972,17 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
   __pyx_v_features_bytes = ((PyObject*)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "bayesclass/BayesNetwork.pyx":31
+  /* "bayesclass/BayesNetwork.pyx":30
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         features_bytes = [x.encode() for x in features]
  *         self.thisptr.fit(X_, y, features_bytes, className.encode())             # <<<<<<<<<<<<<<
  *         return self
  *     def predict(self, X):
  */
-  __pyx_t_8 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 31, __pyx_L1_error)
-  __pyx_t_9 = __pyx_convert_vector_from_py_int(__pyx_v_y); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 31, __pyx_L1_error)
-  __pyx_t_10 = __pyx_convert_vector_from_py_std_3a__3a_string(__pyx_v_features_bytes); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 31, __pyx_L1_error)
-  __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_className, __pyx_n_s_encode); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 31, __pyx_L1_error)
+  __pyx_t_8 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 30, __pyx_L1_error)
+  __pyx_t_9 = __pyx_convert_vector_from_py_int(__pyx_v_y); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 30, __pyx_L1_error)
+  __pyx_t_10 = __pyx_convert_vector_from_py_std_3a__3a_string(__pyx_v_features_bytes); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 30, __pyx_L1_error)
+  __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_className, __pyx_n_s_encode); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 30, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_3);
   __pyx_t_6 = NULL;
   if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_3))) {
@@ -1969,14 +1996,14 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
   }
   __pyx_t_1 = (__pyx_t_6) ? __Pyx_PyObject_CallOneArg(__pyx_t_3, __pyx_t_6) : __Pyx_PyObject_CallNoArg(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_6); __pyx_t_6 = 0;
-  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 31, __pyx_L1_error)
+  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 30, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_11 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 31, __pyx_L1_error)
+  __pyx_t_11 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 30, __pyx_L1_error)
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __pyx_v_self->thisptr->fit(__pyx_t_8, __pyx_t_9, __pyx_t_10, __pyx_t_11);
 
-  /* "bayesclass/BayesNetwork.pyx":32
+  /* "bayesclass/BayesNetwork.pyx":31
  *         features_bytes = [x.encode() for x in features]
  *         self.thisptr.fit(X_, y, features_bytes, className.encode())
  *         return self             # <<<<<<<<<<<<<<
@@ -1988,7 +2015,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
   __pyx_r = ((PyObject *)__pyx_v_self);
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":28
+  /* "bayesclass/BayesNetwork.pyx":27
  *     def __dealloc__(self):
  *         del self.thisptr
  *     def fit(self, X, y, features, className):             # <<<<<<<<<<<<<<
@@ -2015,7 +2042,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_4fit(struct __py
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":33
+/* "bayesclass/BayesNetwork.pyx":32
  *         self.thisptr.fit(X_, y, features_bytes, className.encode())
  *         return self
  *     def predict(self, X):             # <<<<<<<<<<<<<<
@@ -2053,7 +2080,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("predict", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":34
+  /* "bayesclass/BayesNetwork.pyx":33
  *         return self
  *     def predict(self, X):
  *         X_ = [X[:, i] for i in range(X.shape[1])]             # <<<<<<<<<<<<<<
@@ -2061,23 +2088,23 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
  *     def predict_proba(self, X):
  */
   { /* enter inner scope */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 34, __pyx_L5_error)
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 33, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 34, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 33, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 34, __pyx_L5_error)
+    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 33, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 34, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 33, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     if (likely(PyList_CheckExact(__pyx_t_2)) || PyTuple_CheckExact(__pyx_t_2)) {
       __pyx_t_3 = __pyx_t_2; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
       __pyx_t_5 = NULL;
     } else {
-      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 34, __pyx_L5_error)
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 33, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_3);
-      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 34, __pyx_L5_error)
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 33, __pyx_L5_error)
     }
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     for (;;) {
@@ -2085,17 +2112,17 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
         if (likely(PyList_CheckExact(__pyx_t_3))) {
           if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 34, __pyx_L5_error)
+          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 33, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 34, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 33, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         } else {
           if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 34, __pyx_L5_error)
+          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 33, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 34, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 33, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         }
@@ -2105,7 +2132,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
           PyObject* exc_type = PyErr_Occurred();
           if (exc_type) {
             if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
-            else __PYX_ERR(0, 34, __pyx_L5_error)
+            else __PYX_ERR(0, 33, __pyx_L5_error)
           }
           break;
         }
@@ -2113,7 +2140,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
       }
       __Pyx_XDECREF_SET(__pyx_8genexpr2__pyx_v_i, __pyx_t_2);
       __pyx_t_2 = 0;
-      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 34, __pyx_L5_error)
+      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 33, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_INCREF(__pyx_slice_);
       __Pyx_GIVEREF(__pyx_slice_);
@@ -2121,10 +2148,10 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
       __Pyx_INCREF(__pyx_8genexpr2__pyx_v_i);
       __Pyx_GIVEREF(__pyx_8genexpr2__pyx_v_i);
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_8genexpr2__pyx_v_i);
-      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 34, __pyx_L5_error)
+      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 33, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_6);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 34, __pyx_L5_error)
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 33, __pyx_L5_error)
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     }
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -2138,7 +2165,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
   __pyx_v_X_ = ((PyObject*)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "bayesclass/BayesNetwork.pyx":35
+  /* "bayesclass/BayesNetwork.pyx":34
  *     def predict(self, X):
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.predict(X_)             # <<<<<<<<<<<<<<
@@ -2146,14 +2173,14 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_7 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 35, __pyx_L1_error)
-  __pyx_t_1 = __pyx_convert_vector_to_py_int(__pyx_v_self->thisptr->predict(__pyx_t_7)); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 35, __pyx_L1_error)
+  __pyx_t_7 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 34, __pyx_L1_error)
+  __pyx_t_1 = __pyx_convert_vector_to_py_int(__pyx_v_self->thisptr->predict(__pyx_t_7)); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 34, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":33
+  /* "bayesclass/BayesNetwork.pyx":32
  *         self.thisptr.fit(X_, y, features_bytes, className.encode())
  *         return self
  *     def predict(self, X):             # <<<<<<<<<<<<<<
@@ -2177,7 +2204,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_6predict(struct
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":36
+/* "bayesclass/BayesNetwork.pyx":35
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.predict(X_)
  *     def predict_proba(self, X):             # <<<<<<<<<<<<<<
@@ -2215,7 +2242,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("predict_proba", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":37
+  /* "bayesclass/BayesNetwork.pyx":36
  *         return self.thisptr.predict(X_)
  *     def predict_proba(self, X):
  *         X_ = [X[:, i] for i in range(X.shape[1])]             # <<<<<<<<<<<<<<
@@ -2223,23 +2250,23 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
  *     def score(self, X, y):
  */
   { /* enter inner scope */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 37, __pyx_L5_error)
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 36, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 37, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 36, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 37, __pyx_L5_error)
+    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 36, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 37, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 36, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     if (likely(PyList_CheckExact(__pyx_t_2)) || PyTuple_CheckExact(__pyx_t_2)) {
       __pyx_t_3 = __pyx_t_2; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
       __pyx_t_5 = NULL;
     } else {
-      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 37, __pyx_L5_error)
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 36, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_3);
-      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 37, __pyx_L5_error)
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 36, __pyx_L5_error)
     }
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     for (;;) {
@@ -2247,17 +2274,17 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
         if (likely(PyList_CheckExact(__pyx_t_3))) {
           if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 37, __pyx_L5_error)
+          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 36, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 37, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 36, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         } else {
           if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 37, __pyx_L5_error)
+          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 36, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 37, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 36, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         }
@@ -2267,7 +2294,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
           PyObject* exc_type = PyErr_Occurred();
           if (exc_type) {
             if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
-            else __PYX_ERR(0, 37, __pyx_L5_error)
+            else __PYX_ERR(0, 36, __pyx_L5_error)
           }
           break;
         }
@@ -2275,7 +2302,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
       }
       __Pyx_XDECREF_SET(__pyx_8genexpr3__pyx_v_i, __pyx_t_2);
       __pyx_t_2 = 0;
-      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 37, __pyx_L5_error)
+      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 36, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_INCREF(__pyx_slice_);
       __Pyx_GIVEREF(__pyx_slice_);
@@ -2283,10 +2310,10 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
       __Pyx_INCREF(__pyx_8genexpr3__pyx_v_i);
       __Pyx_GIVEREF(__pyx_8genexpr3__pyx_v_i);
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_8genexpr3__pyx_v_i);
-      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 37, __pyx_L5_error)
+      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 36, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_6);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 37, __pyx_L5_error)
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 36, __pyx_L5_error)
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     }
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -2300,7 +2327,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
   __pyx_v_X_ = ((PyObject*)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "bayesclass/BayesNetwork.pyx":38
+  /* "bayesclass/BayesNetwork.pyx":37
  *     def predict_proba(self, X):
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.predict_proba(X_)             # <<<<<<<<<<<<<<
@@ -2308,14 +2335,14 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_7 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 38, __pyx_L1_error)
-  __pyx_t_1 = __pyx_convert_vector_to_py_std_3a__3a_vector_3c_double_3e___(__pyx_v_self->thisptr->predict_proba(__pyx_t_7)); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 38, __pyx_L1_error)
+  __pyx_t_7 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 37, __pyx_L1_error)
+  __pyx_t_1 = __pyx_convert_vector_to_py_std_3a__3a_vector_3c_double_3e___(__pyx_v_self->thisptr->predict_proba(__pyx_t_7)); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 37, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":36
+  /* "bayesclass/BayesNetwork.pyx":35
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.predict(X_)
  *     def predict_proba(self, X):             # <<<<<<<<<<<<<<
@@ -2339,7 +2366,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_8predict_proba(s
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":39
+/* "bayesclass/BayesNetwork.pyx":38
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.predict_proba(X_)
  *     def score(self, X, y):             # <<<<<<<<<<<<<<
@@ -2381,11 +2408,11 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_11score(PyObject
         case  1:
         if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_y)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("score", 1, 2, 2, 1); __PYX_ERR(0, 39, __pyx_L3_error)
+          __Pyx_RaiseArgtupleInvalid("score", 1, 2, 2, 1); __PYX_ERR(0, 38, __pyx_L3_error)
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "score") < 0)) __PYX_ERR(0, 39, __pyx_L3_error)
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "score") < 0)) __PYX_ERR(0, 38, __pyx_L3_error)
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 2) {
       goto __pyx_L5_argtuple_error;
@@ -2398,7 +2425,7 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_11score(PyObject
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("score", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 39, __pyx_L3_error)
+  __Pyx_RaiseArgtupleInvalid("score", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 38, __pyx_L3_error)
   __pyx_L3_error:;
   __Pyx_AddTraceback("bayesclass.BayesNet.BayesNetwork.score", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
@@ -2429,7 +2456,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("score", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":40
+  /* "bayesclass/BayesNetwork.pyx":39
  *         return self.thisptr.predict_proba(X_)
  *     def score(self, X, y):
  *         X_ = [X[:, i] for i in range(X.shape[1])]             # <<<<<<<<<<<<<<
@@ -2437,23 +2464,23 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
  *     def addNode(self, name, states):
  */
   { /* enter inner scope */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 40, __pyx_L5_error)
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 39, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 40, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 39, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 40, __pyx_L5_error)
+    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 39, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 40, __pyx_L5_error)
+    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 39, __pyx_L5_error)
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     if (likely(PyList_CheckExact(__pyx_t_2)) || PyTuple_CheckExact(__pyx_t_2)) {
       __pyx_t_3 = __pyx_t_2; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
       __pyx_t_5 = NULL;
     } else {
-      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 40, __pyx_L5_error)
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 39, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_3);
-      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 40, __pyx_L5_error)
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 39, __pyx_L5_error)
     }
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     for (;;) {
@@ -2461,17 +2488,17 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
         if (likely(PyList_CheckExact(__pyx_t_3))) {
           if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 40, __pyx_L5_error)
+          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 39, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 40, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 39, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         } else {
           if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
           #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
-          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 40, __pyx_L5_error)
+          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 39, __pyx_L5_error)
           #else
-          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 40, __pyx_L5_error)
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 39, __pyx_L5_error)
           __Pyx_GOTREF(__pyx_t_2);
           #endif
         }
@@ -2481,7 +2508,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
           PyObject* exc_type = PyErr_Occurred();
           if (exc_type) {
             if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
-            else __PYX_ERR(0, 40, __pyx_L5_error)
+            else __PYX_ERR(0, 39, __pyx_L5_error)
           }
           break;
         }
@@ -2489,7 +2516,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
       }
       __Pyx_XDECREF_SET(__pyx_8genexpr4__pyx_v_i, __pyx_t_2);
       __pyx_t_2 = 0;
-      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 40, __pyx_L5_error)
+      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 39, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_2);
       __Pyx_INCREF(__pyx_slice_);
       __Pyx_GIVEREF(__pyx_slice_);
@@ -2497,10 +2524,10 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
       __Pyx_INCREF(__pyx_8genexpr4__pyx_v_i);
       __Pyx_GIVEREF(__pyx_8genexpr4__pyx_v_i);
       PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_8genexpr4__pyx_v_i);
-      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 40, __pyx_L5_error)
+      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 39, __pyx_L5_error)
       __Pyx_GOTREF(__pyx_t_6);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 40, __pyx_L5_error)
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 39, __pyx_L5_error)
       __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
     }
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
@@ -2514,7 +2541,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
   __pyx_v_X_ = ((PyObject*)__pyx_t_1);
   __pyx_t_1 = 0;
 
-  /* "bayesclass/BayesNetwork.pyx":41
+  /* "bayesclass/BayesNetwork.pyx":40
  *     def score(self, X, y):
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.score(X_, y)             # <<<<<<<<<<<<<<
@@ -2522,15 +2549,15 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
  *         self.thisptr.addNode(str.encode(name), states)
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_7 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 41, __pyx_L1_error)
-  __pyx_t_8 = __pyx_convert_vector_from_py_int(__pyx_v_y); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 41, __pyx_L1_error)
-  __pyx_t_1 = PyFloat_FromDouble(__pyx_v_self->thisptr->score(__pyx_t_7, __pyx_t_8)); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 41, __pyx_L1_error)
+  __pyx_t_7 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 40, __pyx_L1_error)
+  __pyx_t_8 = __pyx_convert_vector_from_py_int(__pyx_v_y); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 40, __pyx_L1_error)
+  __pyx_t_1 = PyFloat_FromDouble(__pyx_v_self->thisptr->score(__pyx_t_7, __pyx_t_8)); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 40, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":39
+  /* "bayesclass/BayesNetwork.pyx":38
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.predict_proba(X_)
  *     def score(self, X, y):             # <<<<<<<<<<<<<<
@@ -2554,7 +2581,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_10score(struct _
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":42
+/* "bayesclass/BayesNetwork.pyx":41
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.score(X_, y)
  *     def addNode(self, name, states):             # <<<<<<<<<<<<<<
@@ -2596,11 +2623,11 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_13addNode(PyObje
         case  1:
         if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_states)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("addNode", 1, 2, 2, 1); __PYX_ERR(0, 42, __pyx_L3_error)
+          __Pyx_RaiseArgtupleInvalid("addNode", 1, 2, 2, 1); __PYX_ERR(0, 41, __pyx_L3_error)
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "addNode") < 0)) __PYX_ERR(0, 42, __pyx_L3_error)
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "addNode") < 0)) __PYX_ERR(0, 41, __pyx_L3_error)
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 2) {
       goto __pyx_L5_argtuple_error;
@@ -2613,7 +2640,7 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_13addNode(PyObje
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("addNode", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 42, __pyx_L3_error)
+  __Pyx_RaiseArgtupleInvalid("addNode", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 41, __pyx_L3_error)
   __pyx_L3_error:;
   __Pyx_AddTraceback("bayesclass.BayesNet.BayesNetwork.addNode", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
@@ -2639,14 +2666,14 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_12addNode(struct
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("addNode", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":43
+  /* "bayesclass/BayesNetwork.pyx":42
  *         return self.thisptr.score(X_, y)
  *     def addNode(self, name, states):
  *         self.thisptr.addNode(str.encode(name), states)             # <<<<<<<<<<<<<<
  *     def addEdge(self, source, destination):
  *         self.thisptr.addEdge(str.encode(source), str.encode(destination))
  */
-  __pyx_t_2 = __Pyx_PyObject_GetAttrStr(((PyObject *)(&PyUnicode_Type)), __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 43, __pyx_L1_error)
+  __pyx_t_2 = __Pyx_PyObject_GetAttrStr(((PyObject *)(&PyUnicode_Type)), __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 42, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_t_3 = NULL;
   if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_2))) {
@@ -2660,15 +2687,15 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_12addNode(struct
   }
   __pyx_t_1 = (__pyx_t_3) ? __Pyx_PyObject_Call2Args(__pyx_t_2, __pyx_t_3, __pyx_v_name) : __Pyx_PyObject_CallOneArg(__pyx_t_2, __pyx_v_name);
   __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
-  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 43, __pyx_L1_error)
+  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 42, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_4 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 43, __pyx_L1_error)
+  __pyx_t_4 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 42, __pyx_L1_error)
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_5 = __Pyx_PyInt_As_int(__pyx_v_states); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 43, __pyx_L1_error)
+  __pyx_t_5 = __Pyx_PyInt_As_int(__pyx_v_states); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 42, __pyx_L1_error)
   __pyx_v_self->thisptr->addNode(__pyx_t_4, __pyx_t_5);
 
-  /* "bayesclass/BayesNetwork.pyx":42
+  /* "bayesclass/BayesNetwork.pyx":41
  *         X_ = [X[:, i] for i in range(X.shape[1])]
  *         return self.thisptr.score(X_, y)
  *     def addNode(self, name, states):             # <<<<<<<<<<<<<<
@@ -2691,7 +2718,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_12addNode(struct
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":44
+/* "bayesclass/BayesNetwork.pyx":43
  *     def addNode(self, name, states):
  *         self.thisptr.addNode(str.encode(name), states)
  *     def addEdge(self, source, destination):             # <<<<<<<<<<<<<<
@@ -2733,11 +2760,11 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_15addEdge(PyObje
         case  1:
         if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_destination)) != 0)) kw_args--;
         else {
-          __Pyx_RaiseArgtupleInvalid("addEdge", 1, 2, 2, 1); __PYX_ERR(0, 44, __pyx_L3_error)
+          __Pyx_RaiseArgtupleInvalid("addEdge", 1, 2, 2, 1); __PYX_ERR(0, 43, __pyx_L3_error)
         }
       }
       if (unlikely(kw_args > 0)) {
-        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "addEdge") < 0)) __PYX_ERR(0, 44, __pyx_L3_error)
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "addEdge") < 0)) __PYX_ERR(0, 43, __pyx_L3_error)
       }
     } else if (PyTuple_GET_SIZE(__pyx_args) != 2) {
       goto __pyx_L5_argtuple_error;
@@ -2750,7 +2777,7 @@ static PyObject *__pyx_pw_10bayesclass_8BayesNet_12BayesNetwork_15addEdge(PyObje
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("addEdge", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 44, __pyx_L3_error)
+  __Pyx_RaiseArgtupleInvalid("addEdge", 1, 2, 2, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 43, __pyx_L3_error)
   __pyx_L3_error:;
   __Pyx_AddTraceback("bayesclass.BayesNet.BayesNetwork.addEdge", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
@@ -2776,14 +2803,14 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_14addEdge(struct
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("addEdge", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":45
+  /* "bayesclass/BayesNetwork.pyx":44
  *         self.thisptr.addNode(str.encode(name), states)
  *     def addEdge(self, source, destination):
  *         self.thisptr.addEdge(str.encode(source), str.encode(destination))             # <<<<<<<<<<<<<<
  *     def getFeatures(self):
  *         res = self.thisptr.getFeatures()
  */
-  __pyx_t_2 = __Pyx_PyObject_GetAttrStr(((PyObject *)(&PyUnicode_Type)), __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 45, __pyx_L1_error)
+  __pyx_t_2 = __Pyx_PyObject_GetAttrStr(((PyObject *)(&PyUnicode_Type)), __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 44, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_t_3 = NULL;
   if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_2))) {
@@ -2797,12 +2824,12 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_14addEdge(struct
   }
   __pyx_t_1 = (__pyx_t_3) ? __Pyx_PyObject_Call2Args(__pyx_t_2, __pyx_t_3, __pyx_v_source) : __Pyx_PyObject_CallOneArg(__pyx_t_2, __pyx_v_source);
   __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
-  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 45, __pyx_L1_error)
+  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 44, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_4 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 45, __pyx_L1_error)
+  __pyx_t_4 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 44, __pyx_L1_error)
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_2 = __Pyx_PyObject_GetAttrStr(((PyObject *)(&PyUnicode_Type)), __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 45, __pyx_L1_error)
+  __pyx_t_2 = __Pyx_PyObject_GetAttrStr(((PyObject *)(&PyUnicode_Type)), __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 44, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_t_3 = NULL;
   if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_2))) {
@@ -2816,14 +2843,14 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_14addEdge(struct
   }
   __pyx_t_1 = (__pyx_t_3) ? __Pyx_PyObject_Call2Args(__pyx_t_2, __pyx_t_3, __pyx_v_destination) : __Pyx_PyObject_CallOneArg(__pyx_t_2, __pyx_v_destination);
   __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
-  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 45, __pyx_L1_error)
+  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 44, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_5 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 45, __pyx_L1_error)
+  __pyx_t_5 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 44, __pyx_L1_error)
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __pyx_v_self->thisptr->addEdge(__pyx_t_4, __pyx_t_5);
 
-  /* "bayesclass/BayesNetwork.pyx":44
+  /* "bayesclass/BayesNetwork.pyx":43
  *     def addNode(self, name, states):
  *         self.thisptr.addNode(str.encode(name), states)
  *     def addEdge(self, source, destination):             # <<<<<<<<<<<<<<
@@ -2846,7 +2873,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_14addEdge(struct
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":46
+/* "bayesclass/BayesNetwork.pyx":45
  *     def addEdge(self, source, destination):
  *         self.thisptr.addEdge(str.encode(source), str.encode(destination))
  *     def getFeatures(self):             # <<<<<<<<<<<<<<
@@ -2881,7 +2908,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_16getFeatures(st
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("getFeatures", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":47
+  /* "bayesclass/BayesNetwork.pyx":46
  *         self.thisptr.addEdge(str.encode(source), str.encode(destination))
  *     def getFeatures(self):
  *         res = self.thisptr.getFeatures()             # <<<<<<<<<<<<<<
@@ -2890,7 +2917,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_16getFeatures(st
  */
   __pyx_v_res = __pyx_v_self->thisptr->getFeatures();
 
-  /* "bayesclass/BayesNetwork.pyx":48
+  /* "bayesclass/BayesNetwork.pyx":47
  *     def getFeatures(self):
  *         res = self.thisptr.getFeatures()
  *         return [x.decode() for x in res]             # <<<<<<<<<<<<<<
@@ -2899,7 +2926,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_16getFeatures(st
  */
   __Pyx_XDECREF(__pyx_r);
   { /* enter inner scope */
-    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 48, __pyx_L1_error)
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 47, __pyx_L1_error)
     __Pyx_GOTREF(__pyx_t_1);
     __pyx_t_2 = __pyx_v_res.begin();
     for (;;) {
@@ -2907,9 +2934,9 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_16getFeatures(st
       __pyx_t_3 = *__pyx_t_2;
       ++__pyx_t_2;
       __pyx_8genexpr5__pyx_v_x = __pyx_t_3;
-      __pyx_t_4 = __Pyx_decode_cpp_string(__pyx_8genexpr5__pyx_v_x, 0, PY_SSIZE_T_MAX, NULL, NULL, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 48, __pyx_L1_error)
+      __pyx_t_4 = __Pyx_decode_cpp_string(__pyx_8genexpr5__pyx_v_x, 0, PY_SSIZE_T_MAX, NULL, NULL, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 47, __pyx_L1_error)
       __Pyx_GOTREF(__pyx_t_4);
-      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_4))) __PYX_ERR(0, 48, __pyx_L1_error)
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_4))) __PYX_ERR(0, 47, __pyx_L1_error)
       __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     }
   } /* exit inner scope */
@@ -2917,7 +2944,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_16getFeatures(st
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":46
+  /* "bayesclass/BayesNetwork.pyx":45
  *     def addEdge(self, source, destination):
  *         self.thisptr.addEdge(str.encode(source), str.encode(destination))
  *     def getFeatures(self):             # <<<<<<<<<<<<<<
@@ -2937,7 +2964,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_16getFeatures(st
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":49
+/* "bayesclass/BayesNetwork.pyx":48
  *         res = self.thisptr.getFeatures()
  *         return [x.decode() for x in res]
  *     def getStates(self):             # <<<<<<<<<<<<<<
@@ -2967,7 +2994,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_18getStates(stru
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("getStates", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":50
+  /* "bayesclass/BayesNetwork.pyx":49
  *         return [x.decode() for x in res]
  *     def getStates(self):
  *         return self.thisptr.getStates()             # <<<<<<<<<<<<<<
@@ -2975,13 +3002,13 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_18getStates(stru
  *         return self.thisptr.getClassName().decode()
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __Pyx_PyInt_From_int(__pyx_v_self->thisptr->getStates()); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 50, __pyx_L1_error)
+  __pyx_t_1 = __Pyx_PyInt_From_int(__pyx_v_self->thisptr->getStates()); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 49, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":49
+  /* "bayesclass/BayesNetwork.pyx":48
  *         res = self.thisptr.getFeatures()
  *         return [x.decode() for x in res]
  *     def getStates(self):             # <<<<<<<<<<<<<<
@@ -3000,7 +3027,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_18getStates(stru
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":51
+/* "bayesclass/BayesNetwork.pyx":50
  *     def getStates(self):
  *         return self.thisptr.getStates()
  *     def getClassName(self):             # <<<<<<<<<<<<<<
@@ -3030,7 +3057,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_20getClassName(s
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("getClassName", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":52
+  /* "bayesclass/BayesNetwork.pyx":51
  *         return self.thisptr.getStates()
  *     def getClassName(self):
  *         return self.thisptr.getClassName().decode()             # <<<<<<<<<<<<<<
@@ -3038,13 +3065,13 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_20getClassName(s
  *         return self.thisptr.getClassNumStates()
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __Pyx_decode_cpp_string(__pyx_v_self->thisptr->getClassName(), 0, PY_SSIZE_T_MAX, NULL, NULL, NULL); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 52, __pyx_L1_error)
+  __pyx_t_1 = __Pyx_decode_cpp_string(__pyx_v_self->thisptr->getClassName(), 0, PY_SSIZE_T_MAX, NULL, NULL, NULL); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 51, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":51
+  /* "bayesclass/BayesNetwork.pyx":50
  *     def getStates(self):
  *         return self.thisptr.getStates()
  *     def getClassName(self):             # <<<<<<<<<<<<<<
@@ -3063,7 +3090,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_20getClassName(s
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":53
+/* "bayesclass/BayesNetwork.pyx":52
  *     def getClassName(self):
  *         return self.thisptr.getClassName().decode()
  *     def getClassNumStates(self):             # <<<<<<<<<<<<<<
@@ -3093,7 +3120,7 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_22getClassNumSta
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("getClassNumStates", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":54
+  /* "bayesclass/BayesNetwork.pyx":53
  *         return self.thisptr.getClassName().decode()
  *     def getClassNumStates(self):
  *         return self.thisptr.getClassNumStates()             # <<<<<<<<<<<<<<
@@ -3101,13 +3128,13 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_22getClassNumSta
  *         return (BayesNetwork, ())
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __Pyx_PyInt_From_int(__pyx_v_self->thisptr->getClassNumStates()); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 54, __pyx_L1_error)
+  __pyx_t_1 = __Pyx_PyInt_From_int(__pyx_v_self->thisptr->getClassNumStates()); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 53, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":53
+  /* "bayesclass/BayesNetwork.pyx":52
  *     def getClassName(self):
  *         return self.thisptr.getClassName().decode()
  *     def getClassNumStates(self):             # <<<<<<<<<<<<<<
@@ -3126,11 +3153,12 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_22getClassNumSta
   return __pyx_r;
 }
 
-/* "bayesclass/BayesNetwork.pyx":55
+/* "bayesclass/BayesNetwork.pyx":54
  *     def getClassNumStates(self):
  *         return self.thisptr.getClassNumStates()
  *     def __reduce__(self):             # <<<<<<<<<<<<<<
  *         return (BayesNetwork, ())
+ * 
  */
 
 /* Python wrapper */
@@ -3155,13 +3183,15 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_24__reduce__(CYT
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__reduce__", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":56
+  /* "bayesclass/BayesNetwork.pyx":55
  *         return self.thisptr.getClassNumStates()
  *     def __reduce__(self):
  *         return (BayesNetwork, ())             # <<<<<<<<<<<<<<
+ * 
+ * cdef extern from "Metrics.hpp" namespace "bayesnet":
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 56, __pyx_L1_error)
+  __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 55, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_INCREF(((PyObject *)__pyx_ptype_10bayesclass_8BayesNet_BayesNetwork));
   __Pyx_GIVEREF(((PyObject *)__pyx_ptype_10bayesclass_8BayesNet_BayesNetwork));
@@ -3173,11 +3203,12 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_24__reduce__(CYT
   __pyx_t_1 = 0;
   goto __pyx_L0;
 
-  /* "bayesclass/BayesNetwork.pyx":55
+  /* "bayesclass/BayesNetwork.pyx":54
  *     def getClassNumStates(self):
  *         return self.thisptr.getClassNumStates()
  *     def __reduce__(self):             # <<<<<<<<<<<<<<
  *         return (BayesNetwork, ())
+ * 
  */
 
   /* function exit code */
@@ -3191,6 +3222,607 @@ static PyObject *__pyx_pf_10bayesclass_8BayesNet_12BayesNetwork_24__reduce__(CYT
   return __pyx_r;
 }
 
+/* "bayesclass/BayesNetwork.pyx":65
+ * cdef class CMetrics:
+ *     cdef Metrics *thisptr
+ *     def __cinit__(self, X, y, features, className, classStates):             # <<<<<<<<<<<<<<
+ *         X_ = [X[:, i] for i in range(X.shape[1])]
+ *         features_bytes = [x.encode() for x in features]
+ */
+
+/* Python wrapper */
+static int __pyx_pw_10bayesclass_8BayesNet_8CMetrics_1__cinit__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static int __pyx_pw_10bayesclass_8BayesNet_8CMetrics_1__cinit__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+  PyObject *__pyx_v_X = 0;
+  PyObject *__pyx_v_y = 0;
+  PyObject *__pyx_v_features = 0;
+  PyObject *__pyx_v_className = 0;
+  PyObject *__pyx_v_classStates = 0;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__cinit__ (wrapper)", 0);
+  {
+    static PyObject **__pyx_pyargnames[] = {&__pyx_n_s_X,&__pyx_n_s_y,&__pyx_n_s_features,&__pyx_n_s_className,&__pyx_n_s_classStates,0};
+    PyObject* values[5] = {0,0,0,0,0};
+    if (unlikely(__pyx_kwds)) {
+      Py_ssize_t kw_args;
+      const Py_ssize_t pos_args = PyTuple_GET_SIZE(__pyx_args);
+      switch (pos_args) {
+        case  5: values[4] = PyTuple_GET_ITEM(__pyx_args, 4);
+        CYTHON_FALLTHROUGH;
+        case  4: values[3] = PyTuple_GET_ITEM(__pyx_args, 3);
+        CYTHON_FALLTHROUGH;
+        case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
+        CYTHON_FALLTHROUGH;
+        case  2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
+        CYTHON_FALLTHROUGH;
+        case  1: values[0] = PyTuple_GET_ITEM(__pyx_args, 0);
+        CYTHON_FALLTHROUGH;
+        case  0: break;
+        default: goto __pyx_L5_argtuple_error;
+      }
+      kw_args = PyDict_Size(__pyx_kwds);
+      switch (pos_args) {
+        case  0:
+        if (likely((values[0] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_X)) != 0)) kw_args--;
+        else goto __pyx_L5_argtuple_error;
+        CYTHON_FALLTHROUGH;
+        case  1:
+        if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_y)) != 0)) kw_args--;
+        else {
+          __Pyx_RaiseArgtupleInvalid("__cinit__", 1, 5, 5, 1); __PYX_ERR(0, 65, __pyx_L3_error)
+        }
+        CYTHON_FALLTHROUGH;
+        case  2:
+        if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_features)) != 0)) kw_args--;
+        else {
+          __Pyx_RaiseArgtupleInvalid("__cinit__", 1, 5, 5, 2); __PYX_ERR(0, 65, __pyx_L3_error)
+        }
+        CYTHON_FALLTHROUGH;
+        case  3:
+        if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_className)) != 0)) kw_args--;
+        else {
+          __Pyx_RaiseArgtupleInvalid("__cinit__", 1, 5, 5, 3); __PYX_ERR(0, 65, __pyx_L3_error)
+        }
+        CYTHON_FALLTHROUGH;
+        case  4:
+        if (likely((values[4] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_classStates)) != 0)) kw_args--;
+        else {
+          __Pyx_RaiseArgtupleInvalid("__cinit__", 1, 5, 5, 4); __PYX_ERR(0, 65, __pyx_L3_error)
+        }
+      }
+      if (unlikely(kw_args > 0)) {
+        if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "__cinit__") < 0)) __PYX_ERR(0, 65, __pyx_L3_error)
+      }
+    } else if (PyTuple_GET_SIZE(__pyx_args) != 5) {
+      goto __pyx_L5_argtuple_error;
+    } else {
+      values[0] = PyTuple_GET_ITEM(__pyx_args, 0);
+      values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
+      values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
+      values[3] = PyTuple_GET_ITEM(__pyx_args, 3);
+      values[4] = PyTuple_GET_ITEM(__pyx_args, 4);
+    }
+    __pyx_v_X = values[0];
+    __pyx_v_y = values[1];
+    __pyx_v_features = values[2];
+    __pyx_v_className = values[3];
+    __pyx_v_classStates = values[4];
+  }
+  goto __pyx_L4_argument_unpacking_done;
+  __pyx_L5_argtuple_error:;
+  __Pyx_RaiseArgtupleInvalid("__cinit__", 1, 5, 5, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 65, __pyx_L3_error)
+  __pyx_L3_error:;
+  __Pyx_AddTraceback("bayesclass.BayesNet.CMetrics.__cinit__", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __Pyx_RefNannyFinishContext();
+  return -1;
+  __pyx_L4_argument_unpacking_done:;
+  __pyx_r = __pyx_pf_10bayesclass_8BayesNet_8CMetrics___cinit__(((struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *)__pyx_v_self), __pyx_v_X, __pyx_v_y, __pyx_v_features, __pyx_v_className, __pyx_v_classStates);
+
+  /* function exit code */
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_10bayesclass_8BayesNet_8CMetrics___cinit__(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self, PyObject *__pyx_v_X, PyObject *__pyx_v_y, PyObject *__pyx_v_features, PyObject *__pyx_v_className, PyObject *__pyx_v_classStates) {
+  PyObject *__pyx_v_X_ = NULL;
+  PyObject *__pyx_v_features_bytes = NULL;
+  PyObject *__pyx_8genexpr6__pyx_v_i = NULL;
+  PyObject *__pyx_8genexpr7__pyx_v_x = NULL;
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  PyObject *__pyx_t_1 = NULL;
+  PyObject *__pyx_t_2 = NULL;
+  PyObject *__pyx_t_3 = NULL;
+  Py_ssize_t __pyx_t_4;
+  PyObject *(*__pyx_t_5)(PyObject *);
+  PyObject *__pyx_t_6 = NULL;
+  PyObject *__pyx_t_7 = NULL;
+  std::vector<std::vector<int> >  __pyx_t_8;
+  std::vector<int>  __pyx_t_9;
+  std::vector<std::string>  __pyx_t_10;
+  std::string __pyx_t_11;
+  int __pyx_t_12;
+  bayesnet::Metrics *__pyx_t_13;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("__cinit__", 0);
+
+  /* "bayesclass/BayesNetwork.pyx":66
+ *     cdef Metrics *thisptr
+ *     def __cinit__(self, X, y, features, className, classStates):
+ *         X_ = [X[:, i] for i in range(X.shape[1])]             # <<<<<<<<<<<<<<
+ *         features_bytes = [x.encode() for x in features]
+ *         self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)
+ */
+  { /* enter inner scope */
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 66, __pyx_L5_error)
+    __Pyx_GOTREF(__pyx_t_1);
+    __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_v_X, __pyx_n_s_shape); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 66, __pyx_L5_error)
+    __Pyx_GOTREF(__pyx_t_2);
+    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 1); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 66, __pyx_L5_error)
+    __Pyx_GOTREF(__pyx_t_3);
+    __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
+    __pyx_t_2 = __Pyx_PyObject_CallOneArg(__pyx_builtin_range, __pyx_t_3); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 66, __pyx_L5_error)
+    __Pyx_GOTREF(__pyx_t_2);
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    if (likely(PyList_CheckExact(__pyx_t_2)) || PyTuple_CheckExact(__pyx_t_2)) {
+      __pyx_t_3 = __pyx_t_2; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
+      __pyx_t_5 = NULL;
+    } else {
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_t_2); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 66, __pyx_L5_error)
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 66, __pyx_L5_error)
+    }
+    __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
+    for (;;) {
+      if (likely(!__pyx_t_5)) {
+        if (likely(PyList_CheckExact(__pyx_t_3))) {
+          if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
+          #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
+          __pyx_t_2 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 66, __pyx_L5_error)
+          #else
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 66, __pyx_L5_error)
+          __Pyx_GOTREF(__pyx_t_2);
+          #endif
+        } else {
+          if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
+          #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
+          __pyx_t_2 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_2); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 66, __pyx_L5_error)
+          #else
+          __pyx_t_2 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 66, __pyx_L5_error)
+          __Pyx_GOTREF(__pyx_t_2);
+          #endif
+        }
+      } else {
+        __pyx_t_2 = __pyx_t_5(__pyx_t_3);
+        if (unlikely(!__pyx_t_2)) {
+          PyObject* exc_type = PyErr_Occurred();
+          if (exc_type) {
+            if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
+            else __PYX_ERR(0, 66, __pyx_L5_error)
+          }
+          break;
+        }
+        __Pyx_GOTREF(__pyx_t_2);
+      }
+      __Pyx_XDECREF_SET(__pyx_8genexpr6__pyx_v_i, __pyx_t_2);
+      __pyx_t_2 = 0;
+      __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 66, __pyx_L5_error)
+      __Pyx_GOTREF(__pyx_t_2);
+      __Pyx_INCREF(__pyx_slice_);
+      __Pyx_GIVEREF(__pyx_slice_);
+      PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_slice_);
+      __Pyx_INCREF(__pyx_8genexpr6__pyx_v_i);
+      __Pyx_GIVEREF(__pyx_8genexpr6__pyx_v_i);
+      PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_8genexpr6__pyx_v_i);
+      __pyx_t_6 = __Pyx_PyObject_GetItem(__pyx_v_X, __pyx_t_2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 66, __pyx_L5_error)
+      __Pyx_GOTREF(__pyx_t_6);
+      __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 66, __pyx_L5_error)
+      __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+    }
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __Pyx_XDECREF(__pyx_8genexpr6__pyx_v_i); __pyx_8genexpr6__pyx_v_i = 0;
+    goto __pyx_L8_exit_scope;
+    __pyx_L5_error:;
+    __Pyx_XDECREF(__pyx_8genexpr6__pyx_v_i); __pyx_8genexpr6__pyx_v_i = 0;
+    goto __pyx_L1_error;
+    __pyx_L8_exit_scope:;
+  } /* exit inner scope */
+  __pyx_v_X_ = ((PyObject*)__pyx_t_1);
+  __pyx_t_1 = 0;
+
+  /* "bayesclass/BayesNetwork.pyx":67
+ *     def __cinit__(self, X, y, features, className, classStates):
+ *         X_ = [X[:, i] for i in range(X.shape[1])]
+ *         features_bytes = [x.encode() for x in features]             # <<<<<<<<<<<<<<
+ *         self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)
+ *     def __dealloc__(self):
+ */
+  { /* enter inner scope */
+    __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 67, __pyx_L11_error)
+    __Pyx_GOTREF(__pyx_t_1);
+    if (likely(PyList_CheckExact(__pyx_v_features)) || PyTuple_CheckExact(__pyx_v_features)) {
+      __pyx_t_3 = __pyx_v_features; __Pyx_INCREF(__pyx_t_3); __pyx_t_4 = 0;
+      __pyx_t_5 = NULL;
+    } else {
+      __pyx_t_4 = -1; __pyx_t_3 = PyObject_GetIter(__pyx_v_features); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 67, __pyx_L11_error)
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_5 = Py_TYPE(__pyx_t_3)->tp_iternext; if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 67, __pyx_L11_error)
+    }
+    for (;;) {
+      if (likely(!__pyx_t_5)) {
+        if (likely(PyList_CheckExact(__pyx_t_3))) {
+          if (__pyx_t_4 >= PyList_GET_SIZE(__pyx_t_3)) break;
+          #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
+          __pyx_t_6 = PyList_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_6); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 67, __pyx_L11_error)
+          #else
+          __pyx_t_6 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 67, __pyx_L11_error)
+          __Pyx_GOTREF(__pyx_t_6);
+          #endif
+        } else {
+          if (__pyx_t_4 >= PyTuple_GET_SIZE(__pyx_t_3)) break;
+          #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS
+          __pyx_t_6 = PyTuple_GET_ITEM(__pyx_t_3, __pyx_t_4); __Pyx_INCREF(__pyx_t_6); __pyx_t_4++; if (unlikely(0 < 0)) __PYX_ERR(0, 67, __pyx_L11_error)
+          #else
+          __pyx_t_6 = PySequence_ITEM(__pyx_t_3, __pyx_t_4); __pyx_t_4++; if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 67, __pyx_L11_error)
+          __Pyx_GOTREF(__pyx_t_6);
+          #endif
+        }
+      } else {
+        __pyx_t_6 = __pyx_t_5(__pyx_t_3);
+        if (unlikely(!__pyx_t_6)) {
+          PyObject* exc_type = PyErr_Occurred();
+          if (exc_type) {
+            if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear();
+            else __PYX_ERR(0, 67, __pyx_L11_error)
+          }
+          break;
+        }
+        __Pyx_GOTREF(__pyx_t_6);
+      }
+      __Pyx_XDECREF_SET(__pyx_8genexpr7__pyx_v_x, __pyx_t_6);
+      __pyx_t_6 = 0;
+      __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_8genexpr7__pyx_v_x, __pyx_n_s_encode); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 67, __pyx_L11_error)
+      __Pyx_GOTREF(__pyx_t_2);
+      __pyx_t_7 = NULL;
+      if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_2))) {
+        __pyx_t_7 = PyMethod_GET_SELF(__pyx_t_2);
+        if (likely(__pyx_t_7)) {
+          PyObject* function = PyMethod_GET_FUNCTION(__pyx_t_2);
+          __Pyx_INCREF(__pyx_t_7);
+          __Pyx_INCREF(function);
+          __Pyx_DECREF_SET(__pyx_t_2, function);
+        }
+      }
+      __pyx_t_6 = (__pyx_t_7) ? __Pyx_PyObject_CallOneArg(__pyx_t_2, __pyx_t_7) : __Pyx_PyObject_CallNoArg(__pyx_t_2);
+      __Pyx_XDECREF(__pyx_t_7); __pyx_t_7 = 0;
+      if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 67, __pyx_L11_error)
+      __Pyx_GOTREF(__pyx_t_6);
+      __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
+      if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_6))) __PYX_ERR(0, 67, __pyx_L11_error)
+      __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+    }
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __Pyx_XDECREF(__pyx_8genexpr7__pyx_v_x); __pyx_8genexpr7__pyx_v_x = 0;
+    goto __pyx_L14_exit_scope;
+    __pyx_L11_error:;
+    __Pyx_XDECREF(__pyx_8genexpr7__pyx_v_x); __pyx_8genexpr7__pyx_v_x = 0;
+    goto __pyx_L1_error;
+    __pyx_L14_exit_scope:;
+  } /* exit inner scope */
+  __pyx_v_features_bytes = ((PyObject*)__pyx_t_1);
+  __pyx_t_1 = 0;
+
+  /* "bayesclass/BayesNetwork.pyx":68
+ *         X_ = [X[:, i] for i in range(X.shape[1])]
+ *         features_bytes = [x.encode() for x in features]
+ *         self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)             # <<<<<<<<<<<<<<
+ *     def __dealloc__(self):
+ *         del self.thisptr
+ */
+  __pyx_t_8 = __pyx_convert_vector_from_py_std_3a__3a_vector_3c_int_3e___(__pyx_v_X_); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 68, __pyx_L1_error)
+  __pyx_t_9 = __pyx_convert_vector_from_py_int(__pyx_v_y); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 68, __pyx_L1_error)
+  __pyx_t_10 = __pyx_convert_vector_from_py_std_3a__3a_string(__pyx_v_features_bytes); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 68, __pyx_L1_error)
+  __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_className, __pyx_n_s_encode); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 68, __pyx_L1_error)
+  __Pyx_GOTREF(__pyx_t_3);
+  __pyx_t_6 = NULL;
+  if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_3))) {
+    __pyx_t_6 = PyMethod_GET_SELF(__pyx_t_3);
+    if (likely(__pyx_t_6)) {
+      PyObject* function = PyMethod_GET_FUNCTION(__pyx_t_3);
+      __Pyx_INCREF(__pyx_t_6);
+      __Pyx_INCREF(function);
+      __Pyx_DECREF_SET(__pyx_t_3, function);
+    }
+  }
+  __pyx_t_1 = (__pyx_t_6) ? __Pyx_PyObject_CallOneArg(__pyx_t_3, __pyx_t_6) : __Pyx_PyObject_CallNoArg(__pyx_t_3);
+  __Pyx_XDECREF(__pyx_t_6); __pyx_t_6 = 0;
+  if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 68, __pyx_L1_error)
+  __Pyx_GOTREF(__pyx_t_1);
+  __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+  __pyx_t_11 = __pyx_convert_string_from_py_std__in_string(__pyx_t_1); if (unlikely(PyErr_Occurred())) __PYX_ERR(0, 68, __pyx_L1_error)
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  __pyx_t_12 = __Pyx_PyInt_As_int(__pyx_v_classStates); if (unlikely((__pyx_t_12 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 68, __pyx_L1_error)
+  try {
+    __pyx_t_13 = new bayesnet::Metrics(__pyx_t_8, __pyx_t_9, __pyx_t_10, __pyx_t_11, __pyx_t_12);
+  } catch(...) {
+    __Pyx_CppExn2PyErr();
+    __PYX_ERR(0, 68, __pyx_L1_error)
+  }
+  __pyx_v_self->thisptr = __pyx_t_13;
+
+  /* "bayesclass/BayesNetwork.pyx":65
+ * cdef class CMetrics:
+ *     cdef Metrics *thisptr
+ *     def __cinit__(self, X, y, features, className, classStates):             # <<<<<<<<<<<<<<
+ *         X_ = [X[:, i] for i in range(X.shape[1])]
+ *         features_bytes = [x.encode() for x in features]
+ */
+
+  /* function exit code */
+  __pyx_r = 0;
+  goto __pyx_L0;
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
+  __Pyx_XDECREF(__pyx_t_2);
+  __Pyx_XDECREF(__pyx_t_3);
+  __Pyx_XDECREF(__pyx_t_6);
+  __Pyx_XDECREF(__pyx_t_7);
+  __Pyx_AddTraceback("bayesclass.BayesNet.CMetrics.__cinit__", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = -1;
+  __pyx_L0:;
+  __Pyx_XDECREF(__pyx_v_X_);
+  __Pyx_XDECREF(__pyx_v_features_bytes);
+  __Pyx_XDECREF(__pyx_8genexpr6__pyx_v_i);
+  __Pyx_XDECREF(__pyx_8genexpr7__pyx_v_x);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* "bayesclass/BayesNetwork.pyx":69
+ *         features_bytes = [x.encode() for x in features]
+ *         self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)
+ *     def __dealloc__(self):             # <<<<<<<<<<<<<<
+ *         del self.thisptr
+ *     def conditionalEdgeWeights(self):
+ */
+
+/* Python wrapper */
+static void __pyx_pw_10bayesclass_8BayesNet_8CMetrics_3__dealloc__(PyObject *__pyx_v_self); /*proto*/
+static void __pyx_pw_10bayesclass_8BayesNet_8CMetrics_3__dealloc__(PyObject *__pyx_v_self) {
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__dealloc__ (wrapper)", 0);
+  __pyx_pf_10bayesclass_8BayesNet_8CMetrics_2__dealloc__(((struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *)__pyx_v_self));
+
+  /* function exit code */
+  __Pyx_RefNannyFinishContext();
+}
+
+static void __pyx_pf_10bayesclass_8BayesNet_8CMetrics_2__dealloc__(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self) {
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__dealloc__", 0);
+
+  /* "bayesclass/BayesNetwork.pyx":70
+ *         self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)
+ *     def __dealloc__(self):
+ *         del self.thisptr             # <<<<<<<<<<<<<<
+ *     def conditionalEdgeWeights(self):
+ *         return self.thisptr.conditionalEdgeWeights()
+ */
+  delete __pyx_v_self->thisptr;
+
+  /* "bayesclass/BayesNetwork.pyx":69
+ *         features_bytes = [x.encode() for x in features]
+ *         self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)
+ *     def __dealloc__(self):             # <<<<<<<<<<<<<<
+ *         del self.thisptr
+ *     def conditionalEdgeWeights(self):
+ */
+
+  /* function exit code */
+  __Pyx_RefNannyFinishContext();
+}
+
+/* "bayesclass/BayesNetwork.pyx":71
+ *     def __dealloc__(self):
+ *         del self.thisptr
+ *     def conditionalEdgeWeights(self):             # <<<<<<<<<<<<<<
+ *         return self.thisptr.conditionalEdgeWeights()
+ *     def test(self):
+ */
+
+/* Python wrapper */
+static PyObject *__pyx_pw_10bayesclass_8BayesNet_8CMetrics_5conditionalEdgeWeights(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
+static PyObject *__pyx_pw_10bayesclass_8BayesNet_8CMetrics_5conditionalEdgeWeights(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("conditionalEdgeWeights (wrapper)", 0);
+  __pyx_r = __pyx_pf_10bayesclass_8BayesNet_8CMetrics_4conditionalEdgeWeights(((struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *)__pyx_v_self));
+
+  /* function exit code */
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static PyObject *__pyx_pf_10bayesclass_8BayesNet_8CMetrics_4conditionalEdgeWeights(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self) {
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("conditionalEdgeWeights", 0);
+
+  /* "bayesclass/BayesNetwork.pyx":72
+ *         del self.thisptr
+ *     def conditionalEdgeWeights(self):
+ *         return self.thisptr.conditionalEdgeWeights()             # <<<<<<<<<<<<<<
+ *     def test(self):
+ *         return self.thisptr.test()
+ */
+  __Pyx_XDECREF(__pyx_r);
+  __pyx_t_1 = __pyx_convert_vector_to_py_float(__pyx_v_self->thisptr->conditionalEdgeWeights()); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 72, __pyx_L1_error)
+  __Pyx_GOTREF(__pyx_t_1);
+  __pyx_r = __pyx_t_1;
+  __pyx_t_1 = 0;
+  goto __pyx_L0;
+
+  /* "bayesclass/BayesNetwork.pyx":71
+ *     def __dealloc__(self):
+ *         del self.thisptr
+ *     def conditionalEdgeWeights(self):             # <<<<<<<<<<<<<<
+ *         return self.thisptr.conditionalEdgeWeights()
+ *     def test(self):
+ */
+
+  /* function exit code */
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
+  __Pyx_AddTraceback("bayesclass.BayesNet.CMetrics.conditionalEdgeWeights", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = NULL;
+  __pyx_L0:;
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* "bayesclass/BayesNetwork.pyx":73
+ *     def conditionalEdgeWeights(self):
+ *         return self.thisptr.conditionalEdgeWeights()
+ *     def test(self):             # <<<<<<<<<<<<<<
+ *         return self.thisptr.test()
+ *     def __reduce__(self):
+ */
+
+/* Python wrapper */
+static PyObject *__pyx_pw_10bayesclass_8BayesNet_8CMetrics_7test(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
+static PyObject *__pyx_pw_10bayesclass_8BayesNet_8CMetrics_7test(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("test (wrapper)", 0);
+  __pyx_r = __pyx_pf_10bayesclass_8BayesNet_8CMetrics_6test(((struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *)__pyx_v_self));
+
+  /* function exit code */
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static PyObject *__pyx_pf_10bayesclass_8BayesNet_8CMetrics_6test(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self) {
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("test", 0);
+
+  /* "bayesclass/BayesNetwork.pyx":74
+ *         return self.thisptr.conditionalEdgeWeights()
+ *     def test(self):
+ *         return self.thisptr.test()             # <<<<<<<<<<<<<<
+ *     def __reduce__(self):
+ *         return (CMetrics, ())
+ */
+  __Pyx_XDECREF(__pyx_r);
+  __pyx_t_1 = __pyx_convert_vector_to_py_float(__pyx_v_self->thisptr->test()); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 74, __pyx_L1_error)
+  __Pyx_GOTREF(__pyx_t_1);
+  __pyx_r = __pyx_t_1;
+  __pyx_t_1 = 0;
+  goto __pyx_L0;
+
+  /* "bayesclass/BayesNetwork.pyx":73
+ *     def conditionalEdgeWeights(self):
+ *         return self.thisptr.conditionalEdgeWeights()
+ *     def test(self):             # <<<<<<<<<<<<<<
+ *         return self.thisptr.test()
+ *     def __reduce__(self):
+ */
+
+  /* function exit code */
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
+  __Pyx_AddTraceback("bayesclass.BayesNet.CMetrics.test", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = NULL;
+  __pyx_L0:;
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* "bayesclass/BayesNetwork.pyx":75
+ *     def test(self):
+ *         return self.thisptr.test()
+ *     def __reduce__(self):             # <<<<<<<<<<<<<<
+ *         return (CMetrics, ())
+ * 
+ */
+
+/* Python wrapper */
+static PyObject *__pyx_pw_10bayesclass_8BayesNet_8CMetrics_9__reduce__(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
+static PyObject *__pyx_pw_10bayesclass_8BayesNet_8CMetrics_9__reduce__(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
+  PyObject *__pyx_r = 0;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__reduce__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_10bayesclass_8BayesNet_8CMetrics_8__reduce__(((struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *)__pyx_v_self));
+
+  /* function exit code */
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static PyObject *__pyx_pf_10bayesclass_8BayesNet_8CMetrics_8__reduce__(CYTHON_UNUSED struct __pyx_obj_10bayesclass_8BayesNet_CMetrics *__pyx_v_self) {
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("__reduce__", 0);
+
+  /* "bayesclass/BayesNetwork.pyx":76
+ *         return self.thisptr.test()
+ *     def __reduce__(self):
+ *         return (CMetrics, ())             # <<<<<<<<<<<<<<
+ * 
+ */
+  __Pyx_XDECREF(__pyx_r);
+  __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 76, __pyx_L1_error)
+  __Pyx_GOTREF(__pyx_t_1);
+  __Pyx_INCREF(((PyObject *)__pyx_ptype_10bayesclass_8BayesNet_CMetrics));
+  __Pyx_GIVEREF(((PyObject *)__pyx_ptype_10bayesclass_8BayesNet_CMetrics));
+  PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_ptype_10bayesclass_8BayesNet_CMetrics));
+  __Pyx_INCREF(__pyx_empty_tuple);
+  __Pyx_GIVEREF(__pyx_empty_tuple);
+  PyTuple_SET_ITEM(__pyx_t_1, 1, __pyx_empty_tuple);
+  __pyx_r = __pyx_t_1;
+  __pyx_t_1 = 0;
+  goto __pyx_L0;
+
+  /* "bayesclass/BayesNetwork.pyx":75
+ *     def test(self):
+ *         return self.thisptr.test()
+ *     def __reduce__(self):             # <<<<<<<<<<<<<<
+ *         return (CMetrics, ())
+ * 
+ */
+
+  /* function exit code */
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
+  __Pyx_AddTraceback("bayesclass.BayesNet.CMetrics.__reduce__", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = NULL;
+  __pyx_L0:;
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
 /* "vector.from_py":45
  * 
  * @cname("__pyx_convert_vector_from_py_int")
@@ -4042,6 +4674,71 @@ static CYTHON_INLINE PyObject *__pyx_convert_PyByteArray_string_to_py_std__in_st
   return __pyx_r;
 }
 
+/* "vector.to_py":60
+ * 
+ * @cname("__pyx_convert_vector_to_py_float")
+ * cdef object __pyx_convert_vector_to_py_float(vector[X]& v):             # <<<<<<<<<<<<<<
+ *     return [v[i] for i in range(v.size())]
+ * 
+ */
+
+static PyObject *__pyx_convert_vector_to_py_float(const std::vector<float>  &__pyx_v_v) {
+  size_t __pyx_v_i;
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  PyObject *__pyx_t_1 = NULL;
+  size_t __pyx_t_2;
+  size_t __pyx_t_3;
+  size_t __pyx_t_4;
+  PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  __Pyx_RefNannySetupContext("__pyx_convert_vector_to_py_float", 0);
+
+  /* "vector.to_py":61
+ * @cname("__pyx_convert_vector_to_py_float")
+ * cdef object __pyx_convert_vector_to_py_float(vector[X]& v):
+ *     return [v[i] for i in range(v.size())]             # <<<<<<<<<<<<<<
+ * 
+ * 
+ */
+  __Pyx_XDECREF(__pyx_r);
+  __pyx_t_1 = PyList_New(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 61, __pyx_L1_error)
+  __Pyx_GOTREF(__pyx_t_1);
+  __pyx_t_2 = __pyx_v_v.size();
+  __pyx_t_3 = __pyx_t_2;
+  for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) {
+    __pyx_v_i = __pyx_t_4;
+    __pyx_t_5 = PyFloat_FromDouble((__pyx_v_v[__pyx_v_i])); if (unlikely(!__pyx_t_5)) __PYX_ERR(1, 61, __pyx_L1_error)
+    __Pyx_GOTREF(__pyx_t_5);
+    if (unlikely(__Pyx_ListComp_Append(__pyx_t_1, (PyObject*)__pyx_t_5))) __PYX_ERR(1, 61, __pyx_L1_error)
+    __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+  }
+  __pyx_r = __pyx_t_1;
+  __pyx_t_1 = 0;
+  goto __pyx_L0;
+
+  /* "vector.to_py":60
+ * 
+ * @cname("__pyx_convert_vector_to_py_float")
+ * cdef object __pyx_convert_vector_to_py_float(vector[X]& v):             # <<<<<<<<<<<<<<
+ *     return [v[i] for i in range(v.size())]
+ * 
+ */
+
+  /* function exit code */
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
+  __Pyx_XDECREF(__pyx_t_5);
+  __Pyx_AddTraceback("vector.to_py.__pyx_convert_vector_to_py_float", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = 0;
+  __pyx_L0:;
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
 static PyObject *__pyx_tp_new_10bayesclass_8BayesNet_BayesNetwork(PyTypeObject *t, PyObject *a, PyObject *k) {
   PyObject *o;
   if (likely((t->tp_flags & Py_TPFLAGS_IS_ABSTRACT) == 0)) {
@@ -4161,6 +4858,117 @@ static PyTypeObject __pyx_type_10bayesclass_8BayesNet_BayesNetwork = {
   #endif
 };
 
+static PyObject *__pyx_tp_new_10bayesclass_8BayesNet_CMetrics(PyTypeObject *t, PyObject *a, PyObject *k) {
+  PyObject *o;
+  if (likely((t->tp_flags & Py_TPFLAGS_IS_ABSTRACT) == 0)) {
+    o = (*t->tp_alloc)(t, 0);
+  } else {
+    o = (PyObject *) PyBaseObject_Type.tp_new(t, __pyx_empty_tuple, 0);
+  }
+  if (unlikely(!o)) return 0;
+  if (unlikely(__pyx_pw_10bayesclass_8BayesNet_8CMetrics_1__cinit__(o, a, k) < 0)) goto bad;
+  return o;
+  bad:
+  Py_DECREF(o); o = 0;
+  return NULL;
+}
+
+static void __pyx_tp_dealloc_10bayesclass_8BayesNet_CMetrics(PyObject *o) {
+  #if CYTHON_USE_TP_FINALIZE
+  if (unlikely(PyType_HasFeature(Py_TYPE(o), Py_TPFLAGS_HAVE_FINALIZE) && Py_TYPE(o)->tp_finalize) && (!PyType_IS_GC(Py_TYPE(o)) || !__Pyx_PyObject_GC_IsFinalized(o))) {
+    if (PyObject_CallFinalizerFromDealloc(o)) return;
+  }
+  #endif
+  {
+    PyObject *etype, *eval, *etb;
+    PyErr_Fetch(&etype, &eval, &etb);
+    __Pyx_SET_REFCNT(o, Py_REFCNT(o) + 1);
+    __pyx_pw_10bayesclass_8BayesNet_8CMetrics_3__dealloc__(o);
+    __Pyx_SET_REFCNT(o, Py_REFCNT(o) - 1);
+    PyErr_Restore(etype, eval, etb);
+  }
+  (*Py_TYPE(o)->tp_free)(o);
+}
+
+static PyMethodDef __pyx_methods_10bayesclass_8BayesNet_CMetrics[] = {
+  {"conditionalEdgeWeights", (PyCFunction)__pyx_pw_10bayesclass_8BayesNet_8CMetrics_5conditionalEdgeWeights, METH_NOARGS, 0},
+  {"test", (PyCFunction)__pyx_pw_10bayesclass_8BayesNet_8CMetrics_7test, METH_NOARGS, 0},
+  {"__reduce__", (PyCFunction)__pyx_pw_10bayesclass_8BayesNet_8CMetrics_9__reduce__, METH_NOARGS, 0},
+  {0, 0, 0, 0}
+};
+
+static PyTypeObject __pyx_type_10bayesclass_8BayesNet_CMetrics = {
+  PyVarObject_HEAD_INIT(0, 0)
+  "bayesclass.BayesNet.CMetrics", /*tp_name*/
+  sizeof(struct __pyx_obj_10bayesclass_8BayesNet_CMetrics), /*tp_basicsize*/
+  0, /*tp_itemsize*/
+  __pyx_tp_dealloc_10bayesclass_8BayesNet_CMetrics, /*tp_dealloc*/
+  #if PY_VERSION_HEX < 0x030800b4
+  0, /*tp_print*/
+  #endif
+  #if PY_VERSION_HEX >= 0x030800b4
+  0, /*tp_vectorcall_offset*/
+  #endif
+  0, /*tp_getattr*/
+  0, /*tp_setattr*/
+  #if PY_MAJOR_VERSION < 3
+  0, /*tp_compare*/
+  #endif
+  #if PY_MAJOR_VERSION >= 3
+  0, /*tp_as_async*/
+  #endif
+  0, /*tp_repr*/
+  0, /*tp_as_number*/
+  0, /*tp_as_sequence*/
+  0, /*tp_as_mapping*/
+  0, /*tp_hash*/
+  0, /*tp_call*/
+  0, /*tp_str*/
+  0, /*tp_getattro*/
+  0, /*tp_setattro*/
+  0, /*tp_as_buffer*/
+  Py_TPFLAGS_DEFAULT|Py_TPFLAGS_HAVE_VERSION_TAG|Py_TPFLAGS_CHECKTYPES|Py_TPFLAGS_HAVE_NEWBUFFER|Py_TPFLAGS_BASETYPE, /*tp_flags*/
+  0, /*tp_doc*/
+  0, /*tp_traverse*/
+  0, /*tp_clear*/
+  0, /*tp_richcompare*/
+  0, /*tp_weaklistoffset*/
+  0, /*tp_iter*/
+  0, /*tp_iternext*/
+  __pyx_methods_10bayesclass_8BayesNet_CMetrics, /*tp_methods*/
+  0, /*tp_members*/
+  0, /*tp_getset*/
+  0, /*tp_base*/
+  0, /*tp_dict*/
+  0, /*tp_descr_get*/
+  0, /*tp_descr_set*/
+  0, /*tp_dictoffset*/
+  0, /*tp_init*/
+  0, /*tp_alloc*/
+  __pyx_tp_new_10bayesclass_8BayesNet_CMetrics, /*tp_new*/
+  0, /*tp_free*/
+  0, /*tp_is_gc*/
+  0, /*tp_bases*/
+  0, /*tp_mro*/
+  0, /*tp_cache*/
+  0, /*tp_subclasses*/
+  0, /*tp_weaklist*/
+  0, /*tp_del*/
+  0, /*tp_version_tag*/
+  #if PY_VERSION_HEX >= 0x030400a1
+  0, /*tp_finalize*/
+  #endif
+  #if PY_VERSION_HEX >= 0x030800b1 && (!CYTHON_COMPILING_IN_PYPY || PYPY_VERSION_NUM >= 0x07030800)
+  0, /*tp_vectorcall*/
+  #endif
+  #if PY_VERSION_HEX >= 0x030800b4 && PY_VERSION_HEX < 0x03090000
+  0, /*tp_print*/
+  #endif
+  #if CYTHON_COMPILING_IN_PYPY && PY_VERSION_HEX >= 0x03090000 && PY_VERSION_HEX < 0x030a0000
+  0, /*tp_pypy_flags*/
+  #endif
+};
+
 static PyMethodDef __pyx_methods[] = {
   {0, 0, 0, 0}
 };
@@ -4208,8 +5016,10 @@ static struct PyModuleDef __pyx_moduledef = {
 
 static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s_BayesNetwork, __pyx_k_BayesNetwork, sizeof(__pyx_k_BayesNetwork), 0, 0, 1, 1},
+  {&__pyx_n_s_CMetrics, __pyx_k_CMetrics, sizeof(__pyx_k_CMetrics), 0, 0, 1, 1},
   {&__pyx_n_s_X, __pyx_k_X, sizeof(__pyx_k_X), 0, 0, 1, 1},
   {&__pyx_n_s_className, __pyx_k_className, sizeof(__pyx_k_className), 0, 0, 1, 1},
+  {&__pyx_n_s_classStates, __pyx_k_classStates, sizeof(__pyx_k_classStates), 0, 0, 1, 1},
   {&__pyx_n_s_cline_in_traceback, __pyx_k_cline_in_traceback, sizeof(__pyx_k_cline_in_traceback), 0, 0, 1, 1},
   {&__pyx_n_s_destination, __pyx_k_destination, sizeof(__pyx_k_destination), 0, 0, 1, 1},
   {&__pyx_n_s_encode, __pyx_k_encode, sizeof(__pyx_k_encode), 0, 0, 1, 1},
@@ -4228,7 +5038,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {0, 0, 0, 0, 0, 0, 0}
 };
 static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) {
-  __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 29, __pyx_L1_error)
+  __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 28, __pyx_L1_error)
   return 0;
   __pyx_L1_error:;
   return -1;
@@ -4238,14 +5048,14 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) {
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("__Pyx_InitCachedConstants", 0);
 
-  /* "bayesclass/BayesNetwork.pyx":29
+  /* "bayesclass/BayesNetwork.pyx":28
  *         del self.thisptr
  *     def fit(self, X, y, features, className):
  *         X_ = [X[:, i] for i in range(X.shape[1])]             # <<<<<<<<<<<<<<
  *         features_bytes = [x.encode() for x in features]
  *         self.thisptr.fit(X_, y, features_bytes, className.encode())
  */
-  __pyx_slice_ = PySlice_New(Py_None, Py_None, Py_None); if (unlikely(!__pyx_slice_)) __PYX_ERR(0, 29, __pyx_L1_error)
+  __pyx_slice_ = PySlice_New(Py_None, Py_None, Py_None); if (unlikely(!__pyx_slice_)) __PYX_ERR(0, 28, __pyx_L1_error)
   __Pyx_GOTREF(__pyx_slice_);
   __Pyx_GIVEREF(__pyx_slice_);
   __Pyx_RefNannyFinishContext();
@@ -4303,15 +5113,24 @@ static int __Pyx_modinit_type_init_code(void) {
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__Pyx_modinit_type_init_code", 0);
   /*--- Type init code ---*/
-  if (PyType_Ready(&__pyx_type_10bayesclass_8BayesNet_BayesNetwork) < 0) __PYX_ERR(0, 22, __pyx_L1_error)
+  if (PyType_Ready(&__pyx_type_10bayesclass_8BayesNet_BayesNetwork) < 0) __PYX_ERR(0, 21, __pyx_L1_error)
   #if PY_VERSION_HEX < 0x030800B1
   __pyx_type_10bayesclass_8BayesNet_BayesNetwork.tp_print = 0;
   #endif
   if ((CYTHON_USE_TYPE_SLOTS && CYTHON_USE_PYTYPE_LOOKUP) && likely(!__pyx_type_10bayesclass_8BayesNet_BayesNetwork.tp_dictoffset && __pyx_type_10bayesclass_8BayesNet_BayesNetwork.tp_getattro == PyObject_GenericGetAttr)) {
     __pyx_type_10bayesclass_8BayesNet_BayesNetwork.tp_getattro = __Pyx_PyObject_GenericGetAttr;
   }
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_BayesNetwork, (PyObject *)&__pyx_type_10bayesclass_8BayesNet_BayesNetwork) < 0) __PYX_ERR(0, 22, __pyx_L1_error)
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_BayesNetwork, (PyObject *)&__pyx_type_10bayesclass_8BayesNet_BayesNetwork) < 0) __PYX_ERR(0, 21, __pyx_L1_error)
   __pyx_ptype_10bayesclass_8BayesNet_BayesNetwork = &__pyx_type_10bayesclass_8BayesNet_BayesNetwork;
+  if (PyType_Ready(&__pyx_type_10bayesclass_8BayesNet_CMetrics) < 0) __PYX_ERR(0, 63, __pyx_L1_error)
+  #if PY_VERSION_HEX < 0x030800B1
+  __pyx_type_10bayesclass_8BayesNet_CMetrics.tp_print = 0;
+  #endif
+  if ((CYTHON_USE_TYPE_SLOTS && CYTHON_USE_PYTYPE_LOOKUP) && likely(!__pyx_type_10bayesclass_8BayesNet_CMetrics.tp_dictoffset && __pyx_type_10bayesclass_8BayesNet_CMetrics.tp_getattro == PyObject_GenericGetAttr)) {
+    __pyx_type_10bayesclass_8BayesNet_CMetrics.tp_getattro = __Pyx_PyObject_GenericGetAttr;
+  }
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_CMetrics, (PyObject *)&__pyx_type_10bayesclass_8BayesNet_CMetrics) < 0) __PYX_ERR(0, 63, __pyx_L1_error)
+  __pyx_ptype_10bayesclass_8BayesNet_CMetrics = &__pyx_type_10bayesclass_8BayesNet_CMetrics;
   __Pyx_RefNannyFinishContext();
   return 0;
   __pyx_L1_error:;
@@ -4553,11 +5372,11 @@ if (!__Pyx_RefNanny) {
   if (PyDict_SetItem(__pyx_d, __pyx_n_s_test, __pyx_t_1) < 0) __PYX_ERR(0, 1, __pyx_L1_error)
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "string.to_py":55
+  /* "vector.to_py":60
  * 
- * @cname("__pyx_convert_PyByteArray_string_to_py_std__in_string")
- * cdef inline object __pyx_convert_PyByteArray_string_to_py_std__in_string(const string& s):             # <<<<<<<<<<<<<<
- *     return __Pyx_PyByteArray_FromStringAndSize(s.data(), s.size())
+ * @cname("__pyx_convert_vector_to_py_float")
+ * cdef object __pyx_convert_vector_to_py_float(vector[X]& v):             # <<<<<<<<<<<<<<
+ *     return [v[i] for i in range(v.size())]
  * 
  */
 
diff --git a/bayesclass/BayesNetwork.pyx b/bayesclass/BayesNetwork.pyx
index 743ee7c..2fe850e 100644
--- a/bayesclass/BayesNetwork.pyx
+++ b/bayesclass/BayesNetwork.pyx
@@ -3,7 +3,6 @@
 from libcpp.vector cimport vector
 from libcpp.string cimport string
 
-
 cdef extern from "Network.h" namespace "bayesnet":
     cdef cppclass Network:
         Network(float, float) except + 
@@ -54,3 +53,25 @@ cdef class BayesNetwork:
         return self.thisptr.getClassNumStates()
     def __reduce__(self):
         return (BayesNetwork, ())
+
+cdef extern from "Metrics.hpp" namespace "bayesnet":
+    cdef cppclass Metrics:
+        Metrics(vector[vector[int]], vector[int], vector[string]&, string&, int) except +
+        vector[float] conditionalEdgeWeights()
+        vector[float] test()
+
+cdef class CMetrics:
+    cdef Metrics *thisptr
+    def __cinit__(self, X, y, features, className, classStates):
+        X_ = [X[:, i] for i in range(X.shape[1])]
+        features_bytes = [x.encode() for x in features]
+        self.thisptr = new Metrics(X_, y, features_bytes, className.encode(), classStates)
+    def __dealloc__(self):
+        del self.thisptr
+    def conditionalEdgeWeights(self):
+        return self.thisptr.conditionalEdgeWeights()
+    def test(self):
+        return self.thisptr.test()
+    def __reduce__(self):
+        return (CMetrics, ())
+
diff --git a/bayesclass/Metrics.cc b/bayesclass/Metrics.cc
new file mode 100644
index 0000000..dacc0cb
--- /dev/null
+++ b/bayesclass/Metrics.cc
@@ -0,0 +1,114 @@
+#include "Metrics.hpp"
+using namespace std;
+namespace bayesnet {
+    Metrics::Metrics(torch::Tensor& samples, vector<string>& features, string& className, int classNumStates)
+        : samples(samples)
+        , features(features)
+        , className(className)
+        , classNumStates(classNumStates)
+    {
+    }
+    Metrics::Metrics(const vector<vector<int>>& vsamples, const vector<int>& labels, const vector<string>& features, const string& className, const int classNumStates)
+        : features(features)
+        , className(className)
+        , classNumStates(classNumStates)
+    {
+        samples = torch::zeros({ static_cast<int64_t>(vsamples[0].size()), static_cast<int64_t>(vsamples.size() + 1) }, torch::kInt64);
+        for (int i = 0; i < vsamples.size(); ++i) {
+            samples.index_put_({ "...", i }, torch::tensor(vsamples[i], torch::kInt64));
+        }
+        samples.index_put_({ "...", -1 }, torch::tensor(labels, torch::kInt64));
+    }
+    vector<pair<string, string>> Metrics::doCombinations(const vector<string>& source)
+    {
+        vector<pair<string, string>> result;
+        for (int i = 0; i < source.size(); ++i) {
+            string temp = source[i];
+            for (int j = i + 1; j < source.size(); ++j) {
+                result.push_back({ temp, source[j] });
+            }
+        }
+        return result;
+    }
+    vector<float> Metrics::conditionalEdgeWeights()
+    {
+        auto result = vector<double>();
+        auto source = vector<string>(features);
+        source.push_back(className);
+        auto combinations = doCombinations(source);
+        // Compute class prior
+        auto margin = torch::zeros({ classNumStates });
+        for (int value = 0; value < classNumStates; ++value) {
+            auto mask = samples.index({ "...", -1 }) == value;
+            margin[value] = mask.sum().item<float>() / samples.sizes()[0];
+        }
+        for (auto [first, second] : combinations) {
+            int64_t index_first = find(features.begin(), features.end(), first) - features.begin();
+            int64_t index_second = find(features.begin(), features.end(), second) - features.begin();
+            double accumulated = 0;
+            for (int value = 0; value < classNumStates; ++value) {
+                auto mask = samples.index({ "...", -1 }) == value;
+                auto first_dataset = samples.index({ mask, index_first });
+                auto second_dataset = samples.index({ mask, index_second });
+                auto mi = mutualInformation(first_dataset, second_dataset);
+                auto pb = margin[value].item<float>();
+                accumulated += pb * mi;
+            }
+            result.push_back(accumulated);
+        }
+        long n_vars = source.size();
+        auto matrix = torch::zeros({ n_vars, n_vars });
+        auto indices = torch::triu_indices(n_vars, n_vars, 1);
+        for (auto i = 0; i < result.size(); ++i) {
+            auto x = indices[0][i];
+            auto y = indices[1][i];
+            matrix[x][y] = result[i];
+            matrix[y][x] = result[i];
+        }
+        std::vector<float> v(matrix.data_ptr<float>(), matrix.data_ptr<float>() + matrix.numel());
+        return v;
+    }
+    double Metrics::entropy(torch::Tensor& feature)
+    {
+        torch::Tensor counts = feature.bincount();
+        int totalWeight = counts.sum().item<int>();
+        torch::Tensor probs = counts.to(torch::kFloat) / totalWeight;
+        torch::Tensor logProbs = torch::log(probs);
+        torch::Tensor entropy = -probs * logProbs;
+        return entropy.nansum().item<double>();
+    }
+    // H(Y|X) = sum_{x in X} p(x) H(Y|X=x)
+    double Metrics::conditionalEntropy(torch::Tensor& firstFeature, torch::Tensor& secondFeature)
+    {
+        int numSamples = firstFeature.sizes()[0];
+        torch::Tensor featureCounts = secondFeature.bincount();
+        unordered_map<int, unordered_map<int, double>> jointCounts;
+        double totalWeight = 0;
+        for (auto i = 0; i < numSamples; i++) {
+            jointCounts[secondFeature[i].item<int>()][firstFeature[i].item<int>()] += 1;
+            totalWeight += 1;
+        }
+        if (totalWeight == 0)
+            throw invalid_argument("Total weight should not be zero");
+        double entropyValue = 0;
+        for (int value = 0; value < featureCounts.sizes()[0]; ++value) {
+            double p_f = featureCounts[value].item<double>() / totalWeight;
+            double entropy_f = 0;
+            for (auto& [label, jointCount] : jointCounts[value]) {
+                double p_l_f = jointCount / featureCounts[value].item<double>();
+                if (p_l_f > 0) {
+                    entropy_f -= p_l_f * log(p_l_f);
+                } else {
+                    entropy_f = 0;
+                }
+            }
+            entropyValue += p_f * entropy_f;
+        }
+        return entropyValue;
+    }
+    // I(X;Y) = H(Y) - H(Y|X)
+    double Metrics::mutualInformation(torch::Tensor& firstFeature, torch::Tensor& secondFeature)
+    {
+        return entropy(firstFeature) - conditionalEntropy(firstFeature, secondFeature);
+    }
+}
\ No newline at end of file
diff --git a/bayesclass/Metrics.hpp b/bayesclass/Metrics.hpp
new file mode 100644
index 0000000..2587d80
--- /dev/null
+++ b/bayesclass/Metrics.hpp
@@ -0,0 +1,24 @@
+#ifndef BAYESNET_METRICS_H
+#define BAYESNET_METRICS_H
+#include <torch/torch.h>
+#include <vector>
+#include <string>
+using namespace std;
+namespace bayesnet {
+    class Metrics {
+    private:
+        torch::Tensor samples;
+        vector<string> features;
+        string className;
+        int classNumStates;
+        vector<pair<string, string>> doCombinations(const vector<string>&);
+        double entropy(torch::Tensor&);
+        double conditionalEntropy(torch::Tensor&, torch::Tensor&);
+        double mutualInformation(torch::Tensor&, torch::Tensor&);
+    public:
+        Metrics(torch::Tensor&, vector<string>&, string&, int);
+        Metrics(const vector<vector<int>>&, const vector<int>&, const vector<string>&, const string&, const int);
+        vector<float> conditionalEdgeWeights();
+    };
+}
+#endif
\ No newline at end of file
diff --git a/bayesclass/Network.cc b/bayesclass/Network.cc
index 4c039b4..8d336ac 100644
--- a/bayesclass/Network.cc
+++ b/bayesclass/Network.cc
@@ -98,11 +98,14 @@ namespace bayesnet {
         this->className = className;
         dataset.clear();
 
-        // Build dataset
+        // Build dataset & tensor of samples
+        samples = torch::zeros({ static_cast<int64_t>(input_data[0].size()), static_cast<int64_t>(input_data.size() + 1) }, torch::kInt64);
         for (int i = 0; i < featureNames.size(); ++i) {
             dataset[featureNames[i]] = input_data[i];
+            samples.index_put_({ "...", i }, torch::tensor(input_data[i], torch::kInt64));
         }
         dataset[className] = labels;
+        samples.index_put_({ "...", -1 }, torch::tensor(labels, torch::kInt64));
         classNumStates = *max_element(labels.begin(), labels.end()) + 1;
         int maxThreadsRunning = static_cast<int>(std::thread::hardware_concurrency() * maxThreads);
         if (maxThreadsRunning < 1) {
@@ -150,14 +153,14 @@ namespace bayesnet {
         }
     }
 
-    vector<int> Network::predict(const vector<vector<int>>& samples)
+    vector<int> Network::predict(const vector<vector<int>>& tsamples)
     {
         vector<int> predictions;
         vector<int> sample;
-        for (int row = 0; row < samples[0].size(); ++row) {
+        for (int row = 0; row < tsamples[0].size(); ++row) {
             sample.clear();
-            for (int col = 0; col < samples.size(); ++col) {
-                sample.push_back(samples[col][row]);
+            for (int col = 0; col < tsamples.size(); ++col) {
+                sample.push_back(tsamples[col][row]);
             }
             vector<double> classProbabilities = predict_sample(sample);
             // Find the class with the maximum posterior probability
@@ -167,22 +170,22 @@ namespace bayesnet {
         }
         return predictions;
     }
-    vector<vector<double>> Network::predict_proba(const vector<vector<int>>& samples)
+    vector<vector<double>> Network::predict_proba(const vector<vector<int>>& tsamples)
     {
         vector<vector<double>> predictions;
         vector<int> sample;
-        for (int row = 0; row < samples[0].size(); ++row) {
+        for (int row = 0; row < tsamples[0].size(); ++row) {
             sample.clear();
-            for (int col = 0; col < samples.size(); ++col) {
-                sample.push_back(samples[col][row]);
+            for (int col = 0; col < tsamples.size(); ++col) {
+                sample.push_back(tsamples[col][row]);
             }
             predictions.push_back(predict_sample(sample));
         }
         return predictions;
     }
-    double Network::score(const vector<vector<int>>& samples, const vector<int>& labels)
+    double Network::score(const vector<vector<int>>& tsamples, const vector<int>& labels)
     {
-        vector<int> y_pred = predict(samples);
+        vector<int> y_pred = predict(tsamples);
         int correct = 0;
         for (int i = 0; i < y_pred.size(); ++i) {
             if (y_pred[i] == labels[i]) {
@@ -238,4 +241,83 @@ namespace bayesnet {
         }
         return result;
     }
+    double Network::mutual_info(torch::Tensor& first, torch::Tensor& second)
+    {
+        return 1;
+    }
+    torch::Tensor Network::conditionalEdgeWeight()
+    {
+        auto result = vector<double>();
+        auto source = vector<string>(features);
+        source.push_back(className);
+        auto combinations = nodes[className]->combinations(source);
+        auto margin = nodes[className]->getCPT();
+        for (auto [first, second] : combinations) {
+            int64_t index_first = find(features.begin(), features.end(), first) - features.begin();
+            int64_t index_second = find(features.begin(), features.end(), second) - features.begin();
+            double accumulated = 0;
+            for (int value = 0; value < classNumStates; ++value) {
+                auto mask = samples.index({ "...", -1 }) == value;
+                auto first_dataset = samples.index({ mask, index_first });
+                auto second_dataset = samples.index({ mask, index_second });
+                auto mi = mutualInformation(first_dataset, second_dataset);
+                auto pb = margin[value].item<float>();
+                accumulated += pb * mi;
+            }
+            result.push_back(accumulated);
+        }
+        long n_vars = source.size();
+        auto matrix = torch::zeros({ n_vars, n_vars });
+        auto indices = torch::triu_indices(n_vars, n_vars, 1);
+        for (auto i = 0; i < result.size(); ++i) {
+            auto x = indices[0][i];
+            auto y = indices[1][i];
+            matrix[x][y] = result[i];
+            matrix[y][x] = result[i];
+        }
+        return matrix;
+    }
+    double Network::entropy(torch::Tensor& feature)
+    {
+        torch::Tensor counts = feature.bincount();
+        int totalWeight = counts.sum().item<int>();
+        torch::Tensor probs = counts.to(torch::kFloat) / totalWeight;
+        torch::Tensor logProbs = torch::log(probs);
+        torch::Tensor entropy = -probs * logProbs;
+        return entropy.nansum().item<double>();
+    }
+    // H(Y|X) = sum_{x in X} p(x) H(Y|X=x)
+    double Network::conditionalEntropy(torch::Tensor& firstFeature, torch::Tensor& secondFeature)
+    {
+        int numSamples = firstFeature.sizes()[0];
+        torch::Tensor featureCounts = secondFeature.bincount();
+        unordered_map<int, unordered_map<int, double>> jointCounts;
+        double totalWeight = 0;
+        for (auto i = 0; i < numSamples; i++) {
+            jointCounts[secondFeature[i].item<int>()][firstFeature[i].item<int>()] += 1;
+            totalWeight += 1;
+        }
+        if (totalWeight == 0)
+            throw invalid_argument("Total weight should not be zero");
+        double entropyValue = 0;
+        for (int value = 0; value < featureCounts.sizes()[0]; ++value) {
+            double p_f = featureCounts[value].item<double>() / totalWeight;
+            double entropy_f = 0;
+            for (auto& [label, jointCount] : jointCounts[value]) {
+                double p_l_f = jointCount / featureCounts[value].item<double>();
+                if (p_l_f > 0) {
+                    entropy_f -= p_l_f * log(p_l_f);
+                } else {
+                    entropy_f = 0;
+                }
+            }
+            entropyValue += p_f * entropy_f;
+        }
+        return entropyValue;
+    }
+    // I(X;Y) = H(Y) - H(Y|X)
+    double Network::mutualInformation(torch::Tensor& firstFeature, torch::Tensor& secondFeature)
+    {
+        return entropy(firstFeature) - conditionalEntropy(firstFeature, secondFeature);
+    }
 }
diff --git a/bayesclass/Network.h b/bayesclass/Network.h
index 7db04f9..0ba6783 100644
--- a/bayesclass/Network.h
+++ b/bayesclass/Network.h
@@ -19,7 +19,12 @@ namespace bayesnet {
         vector<double> predict_sample(const vector<int>&);
         vector<double> exactInference(map<string, int>&);
         double computeFactor(map<string, int>&);
+        double mutual_info(torch::Tensor&, torch::Tensor&);
+        double entropy(torch::Tensor&);
+        double conditionalEntropy(torch::Tensor&, torch::Tensor&);
+        double mutualInformation(torch::Tensor&, torch::Tensor&);
     public:
+        torch::Tensor samples;
         Network();
         Network(float, int);
         Network(float);
@@ -35,6 +40,8 @@ namespace bayesnet {
         string getClassName();
         void fit(const vector<vector<int>>&, const vector<int>&, const vector<string>&, const string&);
         vector<int> predict(const vector<vector<int>>&);
+        //Computes the conditional edge weight of variable index u and v conditioned on class_node
+        torch::Tensor conditionalEdgeWeight();
         vector<vector<double>> predict_proba(const vector<vector<int>>&);
         double score(const vector<vector<int>>&, const vector<int>&);
         inline string version() { return "0.1.0"; }
diff --git a/bayesclass/Node.cc b/bayesclass/Node.cc
index 075353d..2c5a04d 100644
--- a/bayesclass/Node.cc
+++ b/bayesclass/Node.cc
@@ -57,23 +57,23 @@ namespace bayesnet {
     */
     unsigned Node::minFill()
     {
-        set<string> neighbors;
+        unordered_set<string> neighbors;
         for (auto child : children) {
             neighbors.emplace(child->getName());
         }
         for (auto parent : parents) {
             neighbors.emplace(parent->getName());
         }
-        return combinations(neighbors).size();
+        auto source = vector<string>(neighbors.begin(), neighbors.end());
+        return combinations(source).size();
     }
-    vector<string> Node::combinations(const set<string>& neighbors)
+    vector<pair<string, string>> Node::combinations(const vector<string>& source)
     {
-        vector<string> source(neighbors.begin(), neighbors.end());
-        vector<string> result;
+        vector<pair<string, string>> result;
         for (int i = 0; i < source.size(); ++i) {
             string temp = source[i];
             for (int j = i + 1; j < source.size(); ++j) {
-                result.push_back(temp + source[j]);
+                result.push_back({ temp, source[j] });
             }
         }
         return result;
diff --git a/bayesclass/Node.h b/bayesclass/Node.h
index 817ecb5..c7961aa 100644
--- a/bayesclass/Node.h
+++ b/bayesclass/Node.h
@@ -1,7 +1,7 @@
 #ifndef NODE_H
 #define NODE_H
 #include <torch/torch.h>
-//#include <torch/extension.h>
+#include <unordered_set>
 #include <vector>
 #include <string>
 namespace bayesnet {
@@ -14,8 +14,8 @@ namespace bayesnet {
         int numStates; // number of states of the variable
         torch::Tensor cpTable; // Order of indices is 0-> node variable, 1-> 1st parent, 2-> 2nd parent, ...
         vector<int64_t> dimensions; // dimensions of the cpTable
-        vector<string> combinations(const set<string>&);
     public:
+        vector<pair<string, string>> combinations(const vector<string>&);
         Node(const std::string&, int);
         void addParent(Node*);
         void addChild(Node*);
diff --git a/bayesclass/_version.py b/bayesclass/_version.py
index 485f44a..d3ec452 100644
--- a/bayesclass/_version.py
+++ b/bayesclass/_version.py
@@ -1 +1 @@
-__version__ = "0.1.1"
+__version__ = "0.2.0"
diff --git a/bayesclass/cSelectFeatures.cpp b/bayesclass/cSelectFeatures.cpp
index 877f744..e645610 100644
--- a/bayesclass/cSelectFeatures.cpp
+++ b/bayesclass/cSelectFeatures.cpp
@@ -1,4 +1,4 @@
-/* Generated by Cython 0.29.35 */
+/* Generated by Cython 0.29.36 */
 
 #ifndef PY_SSIZE_T_CLEAN
 #define PY_SSIZE_T_CLEAN
@@ -9,8 +9,8 @@
 #elif PY_VERSION_HEX < 0x02060000 || (0x03000000 <= PY_VERSION_HEX && PY_VERSION_HEX < 0x03030000)
     #error Cython requires Python 2.6+ or Python 3.3+.
 #else
-#define CYTHON_ABI "0_29_35"
-#define CYTHON_HEX_VERSION 0x001D23F0
+#define CYTHON_ABI "0_29_36"
+#define CYTHON_HEX_VERSION 0x001D24F0
 #define CYTHON_FUTURE_DIVISION 1
 #include <stddef.h>
 #ifndef offsetof
@@ -85,7 +85,7 @@
     #define CYTHON_PEP489_MULTI_PHASE_INIT 1
   #endif
   #undef CYTHON_USE_TP_FINALIZE
-  #define CYTHON_USE_TP_FINALIZE 0
+  #define CYTHON_USE_TP_FINALIZE (PY_VERSION_HEX >= 0x030400a1 && PYPY_VERSION_NUM >= 0x07030C00)
   #undef CYTHON_USE_DICT_VERSIONS
   #define CYTHON_USE_DICT_VERSIONS 0
   #undef CYTHON_USE_EXC_INFO_STACK
@@ -383,9 +383,6 @@ class __Pyx_FakeReference {
     T *ptr;
 };
 
-#if CYTHON_COMPILING_IN_PYPY && PY_VERSION_HEX < 0x02070600 && !defined(Py_OptimizeFlag)
-  #define Py_OptimizeFlag 0
-#endif
 #define __PYX_BUILD_PY_SSIZE_T "n"
 #define CYTHON_FORMAT_SSIZE_T "z"
 #if PY_MAJOR_VERSION < 3
@@ -463,6 +460,11 @@ class __Pyx_FakeReference {
 #endif
   #define __Pyx_DefaultClassType PyType_Type
 #endif
+#if PY_VERSION_HEX >= 0x030900F0 && !CYTHON_COMPILING_IN_PYPY
+  #define __Pyx_PyObject_GC_IsFinalized(o) PyObject_GC_IsFinalized(o)
+#else
+  #define __Pyx_PyObject_GC_IsFinalized(o) _PyGC_FINALIZED(o)
+#endif
 #ifndef Py_TPFLAGS_CHECKTYPES
   #define Py_TPFLAGS_CHECKTYPES 0
 #endif
@@ -2601,7 +2603,7 @@ static PyObject *__pyx_tp_new_10bayesclass_17cppSelectFeatures_CSelectKBestWeigh
 
 static void __pyx_tp_dealloc_10bayesclass_17cppSelectFeatures_CSelectKBestWeighted(PyObject *o) {
   #if CYTHON_USE_TP_FINALIZE
-  if (unlikely(PyType_HasFeature(Py_TYPE(o), Py_TPFLAGS_HAVE_FINALIZE) && Py_TYPE(o)->tp_finalize) && (!PyType_IS_GC(Py_TYPE(o)) || !_PyGC_FINALIZED(o))) {
+  if (unlikely(PyType_HasFeature(Py_TYPE(o), Py_TPFLAGS_HAVE_FINALIZE) && Py_TYPE(o)->tp_finalize) && (!PyType_IS_GC(Py_TYPE(o)) || !__Pyx_PyObject_GC_IsFinalized(o))) {
     if (PyObject_CallFinalizerFromDealloc(o)) return;
   }
   #endif
diff --git a/bayesclass/clfs.py b/bayesclass/clfs.py
index e8cfb52..81db853 100644
--- a/bayesclass/clfs.py
+++ b/bayesclass/clfs.py
@@ -16,7 +16,7 @@ from pgmpy.base import DAG
 import matplotlib.pyplot as plt
 from fimdlp.mdlp import FImdlp
 from .cppSelectFeatures import CSelectKBestWeighted
-from .BayesNet import BayesNetwork
+from .BayesNet import BayesNetwork, CMetrics
 from ._version import __version__
 
 
@@ -144,7 +144,7 @@ class BayesBase(BaseEstimator, ClassifierMixin):
         # Store the information needed to build the model
         self.build_dataset()
         # Build the DAG
-        self._build()
+        self._build(kwargs)
         # Train the model
         self._train(kwargs)
         self.fitted_ = True
@@ -153,11 +153,14 @@ class BayesBase(BaseEstimator, ClassifierMixin):
         # Return the classifier
         return self
 
-    def _build(self):
-        """This method should be implemented by the subclasses to
-        build the DAG
-        """
-        ...
+    def _build(self, kwargs):
+        self.model_ = BayesNetwork()
+        features = kwargs["features"]
+        states = kwargs["state_names"]
+        for feature in features:
+            self.model_.addNode(feature, len(states[feature]))
+        class_name = kwargs["class_name"]
+        self.model_.addNode(class_name, max(self.y_) + 1)
 
     def _train(self, kwargs):
         """Build and train a BayesianNetwork from the DAG and the dataset
@@ -178,14 +181,10 @@ class BayesBase(BaseEstimator, ClassifierMixin):
         #     weighted=self.weighted_,
         #     **states,
         # )
-        self.model_ = BayesNetwork()
+
         features = kwargs["features"]
-        states = kwargs["state_names"]
-        for feature in features:
-            self.model_.addNode(feature, len(states[feature]))
         class_name = kwargs["class_name"]
-        self.model_.addNode(class_name, max(self.y_) + 1)
-        for source, destination in self.dag_.edges():
+        for source, destination in self.edges_:
             self.model_.addEdge(source, destination)
         self.model_.fit(self.X_, self.y_, features, class_name)
         self.states_computed_ = self.model_.getStates()
@@ -307,7 +306,7 @@ class TAN(BayesBase):
             raise ValueError("Head index out of range")
         return X, y
 
-    def _build(self):
+    def _build(self, kwargs):
         est = TreeSearch(
             self.dataset_, root_node=self.feature_names_in_[self.head_]
         )
@@ -360,7 +359,7 @@ class KDB(BayesBase):
         ]
         return self._check_params_fit(X, y, expected_args, kwargs)
 
-    def _add_m_edges(self, dag, idx, S_nodes, conditional_weights):
+    def _add_m_edges(self, idx, S_nodes, conditional_weights):
         n_edges = min(self.k, len(S_nodes))
         cond_w = conditional_weights.copy()
         exit_cond = self.k == 0
@@ -369,7 +368,7 @@ class KDB(BayesBase):
             max_minfo = np.argmax(cond_w[idx, :])
             if max_minfo in S_nodes and cond_w[idx, max_minfo] > self.theta:
                 try:
-                    dag.add_edge(
+                    self.add_edge(
                         self.feature_names_in_[max_minfo],
                         self.feature_names_in_[idx],
                     )
@@ -380,7 +379,7 @@ class KDB(BayesBase):
             cond_w[idx, max_minfo] = -1
             exit_cond = num == n_edges or np.all(cond_w[idx, :] <= self.theta)
 
-    def _build(self):
+    def _build(self, kwargs):
         """
         1. For each feature Xi, compute mutual information, I(X;C),
         where C is the class.
@@ -400,14 +399,20 @@ class KDB(BayesBase):
         Compute the conditional probabilility infered by the structure of BN by
         using counts from DB, and output BN.
         """
+        super()._build(kwargs)
         # 1. get the mutual information between each feature and the class
         mutual = mutual_info_classif(self.X_, self.y_, discrete_features=True)
         # 2. symmetric matrix where each element represents I(X, Y| class_node)
-        conditional_weights = TreeSearch(
-            self.dataset_
-        )._get_conditional_weights(
-            self.dataset_, self.class_name_, show_progress=self.show_progress
+        metrics = CMetrics(
+            self.X_,
+            self.y_,
+            self.features_,
+            self.class_name_,
+            self.n_classes_,
         )
+        c_weights = np.array(metrics.conditionalEdgeWeights())
+        n_var = self.n_features_in_ + 1
+        conditional_weights = np.reshape(c_weights, (n_var, n_var))
         '''
         # Step 1: Compute edge weights for a fully connected graph.
         n_vars = len(data.columns)
@@ -442,18 +447,15 @@ class KDB(BayesBase):
         # 3. Let the used variable list, S, be empty.
         S_nodes = []
         # 4. Let the DAG being constructed, BN, begin with a single class node
-        dag = BayesianNetwork()
-        dag.add_node(self.class_name_)  # , state_names=self.classes_)
         # 5. Repeat until S includes all domain features
         # 5.1 Select feature Xmax which is not in S and has the largest value
         for idx in np.argsort(mutual):
             # 5.2 Add a node to BN representing Xmax.
             feature = self.feature_names_in_[idx]
-            dag.add_node(feature)
             # 5.3 Add an arc from C to Xmax in BN.
-            dag.add_edge(self.class_name_, feature)
+            self.edges_.append(self.class_name_, feature)
             # 5.4 Add m = min(lSl,/c) arcs from m distinct features Xj in S
-            self._add_m_edges(dag, idx, S_nodes, conditional_weights)
+            self._add_m_edges(idx, S_nodes, conditional_weights)
             # 5.5 Add Xmax to S.
             S_nodes.append(idx)
         self.dag_ = dag
@@ -851,7 +853,7 @@ class BoostSPODE(BayesBase):
         ]
         return self._check_params_fit(X, y, expected_args, kwargs)
 
-    def _build(self):
+    def _build(self, _):
         class_edges = [(self.class_name_, f) for f in self.feature_names_in_]
         feature_edges = [
             (self.sparent_, f)
diff --git a/setup.py b/setup.py
index f69118e..ac34467 100644
--- a/setup.py
+++ b/setup.py
@@ -32,6 +32,7 @@ setup(
                 "bayesclass/BayesNetwork.pyx",
                 "bayesclass/Network.cc",
                 "bayesclass/Node.cc",
+                "bayesclass/Metrics.cc",
             ],
             include_dirs=include_paths(),
         ),