add kernel hyperparameter subset in gridsearch

analysis_mysql.py

@@ -55,11 +55,24 @@ def report_line(line):
 
 
 def report_footer(agg):
-    print(TextColor.GREEN + f"we have better results {agg['better']:2d} times")
-    print(TextColor.RED + f"we have worse  results {agg['worse']:2d} times")
+    print(
+        TextColor.GREEN
+        + f"we have better results {agg['better']['items']:2d} times"
+    )
+    print(
+        TextColor.RED
+        + f"we have worse  results {agg['worse']['items']:2d} times"
+    )
     color = TextColor.LINE1
     for item in models:
-        print(color + f"{item:10s} used {agg[item]:2d} times")
+        print(
+            color + f"{item:10s} used {agg[item]['items']:2d} times ", end=""
+        )
+        print(
+            color + f"better {agg[item]['better']:2d} times ",
+            end="",
+        )
+        print(color + f"worse {agg[item]['worse']:2d} times ")
         color = (
             TextColor.LINE2 if color == TextColor.LINE1 else TextColor.LINE1
         )
@@ -77,7 +90,10 @@ for item in [
     "better",
     "worse",
 ] + models:
-    agg[item] = 0
+    agg[item] = {}
+    agg[item]["items"] = 0
+    agg[item]["better"] = 0
+    agg[item]["worse"] = 0
 for dataset in dt:
     find_one = False
     line = {"dataset": color + dataset[0]}
@@ -91,13 +107,15 @@ for dataset in dt:
             reference = record[10]
             accuracy = record[5]
             find_one = True
-            agg[model] += 1
+            agg[model]["items"] += 1
             if accuracy > reference:
                 sign = "+"
-                agg["better"] += 1
+                agg["better"]["items"] += 1
+                agg[model]["better"] += 1
             else:
                 sign = "-"
-                agg["worse"] += 1
+                agg["worse"]["items"] += 1
+                agg[model]["worse"] += 1
             item = f"{accuracy:9.7} {sign}"
             line["reference"] = f"{reference:9.7}"
             line[model] = (

experiment.py

@@ -47,6 +47,20 @@ def parse_arguments() -> Tuple[str, str, str, str, str, bool, bool, dict]:
         help="Experiment: {gridsearch, gridbest, crossval, report_grid, "
         "report_cross}",
     )
+    ap.add_argument(
+        "-k",
+        "--kernel",
+        type=str,
+        choices=[
+            "linear",
+            "poly",
+            "rbf",
+            "any",
+        ],
+        required=False,
+        default="any",
+        help="Kernel: {linear, poly, rbf, any} only used in gridsearch",
+    )
     ap.add_argument(
         "-d",
         "--dataset",
@@ -88,6 +102,7 @@ def parse_arguments() -> Tuple[str, str, str, str, str, bool, bool, dict]:
         args.normalize,
         args.standardize,
         args.excludeparams,
+        args.kernel,
     )
 
 
@@ -100,10 +115,15 @@ def parse_arguments() -> Tuple[str, str, str, str, str, bool, bool, dict]:
     normalize,
     standardize,
     exclude_params,
+    kernel,
 ) = parse_arguments()
 
 experiment = Experiment(
-    random_state=1, model=model, host=host, set_of_files=set_of_files
+    random_state=1,
+    model=model,
+    host=host,
+    set_of_files=set_of_files,
+    kernel=kernel,
 )
 if experiment_type[0:6] == "report":
     bd = (

experimentation/Experiments.py

@@ -12,7 +12,12 @@ from .Sets import Datasets
 
 class Experiment:
     def __init__(
-        self, random_state: int, model: str, host: str, set_of_files: str
+        self,
+        random_state: int,
+        model: str,
+        host: str,
+        set_of_files: str,
+        kernel: str,
     ) -> None:
         self._random_state = random_state
         self._model_name = model
@@ -26,6 +31,7 @@ class Experiment:
         # used in gridsearch with ensembles to take best hyperparams of
         # base class or gridsearch these hyperparams as well
         self._base_params = "any"
+        self._kernel = kernel
 
     def set_base_params(self, base_params: str) -> None:
         self._base_params = base_params
@@ -73,6 +79,12 @@ class Experiment:
         """
         hyperparams = Hyperparameters(host=self._host, model=self._model_name)
         model = self._clf.get_model()
+        if self._kernel != "any":
+            # set parameters grid to only one kernel
+            if isinstance(self._clf, Models.Ensemble):
+                self._clf._base_model.select_params(self._kernel)
+            else:
+                self._clf.select_params(self._kernel)
         hyperparameters = self._clf.get_parameters()
         grid_type = "gridsearch"
         if (
@@ -111,7 +123,8 @@ class Experiment:
                 model,
                 return_train_score=True,
                 param_grid=hyperparameters,
-                n_jobs=-1,
+                n_jobs=1,
+                verbose=1,
             )
             start_time = time.time()
             grid_search.fit(X, y)

experimentation/Models.py

@@ -49,35 +49,46 @@ class ModelStree(ModelBase):
         gamma = [1e-1, 1, 1e1]
         max_features = [None, "auto"]
         split_criteria = ["impurity", "max_samples"]
+        self._linear = {
+            "random_state": [self._random_state],
+            "C": C,
+            "max_iter": max_iter,
+            "split_criteria": split_criteria,
+            "max_features": max_features,
+        }
+        self._poly = {
+            "random_state": [self._random_state],
+            "kernel": ["rbf"],
+            "C": C,
+            "gamma": gamma,
+            "max_iter": max_iter,
+            "split_criteria": split_criteria,
+            "max_features": max_features,
+        }
+        self._rbf = {
+            "random_state": [self._random_state],
+            "kernel": ["poly"],
+            "degree": [3, 5],
+            "C": C,
+            "gamma": gamma,
+            "max_iter": max_iter,
+            "split_criteria": split_criteria,
+            "max_features": max_features,
+        }
         self._param_grid = [
-            {
-                "random_state": [self._random_state],
-                "C": C,
-                "max_iter": max_iter,
-                "split_criteria": split_criteria,
-                "max_features": max_features,
-            },
-            {
-                "random_state": [self._random_state],
-                "kernel": ["rbf"],
-                "C": C,
-                "gamma": gamma,
-                "max_iter": max_iter,
-                "split_criteria": split_criteria,
-                "max_features": max_features,
-            },
-            {
-                "random_state": [self._random_state],
-                "kernel": ["poly"],
-                "degree": [3, 5],
-                "C": C,
-                "gamma": gamma,
-                "max_iter": max_iter,
-                "split_criteria": split_criteria,
-                "max_features": max_features,
-            },
+            self._linear,
+            self._poly,
+            self._rbf,
         ]
 
+    def select_params(self, kernel: str) -> None:
+        if kernel == "linear":
+            self._param_grid = [self._linear]
+        elif kernel == "poly":
+            self._param_grid = [self._poly]
+        else:
+            self._param_grid = [self._rbf]
+
 
 class ModelSVC(ModelBase):
     def __init__(self, random_state: Optional[int] = None) -> None: