Add min_length as percentage of # samples

src/fimdlp/mdlp.py

@@ -98,6 +98,11 @@ class FImdlp(TransformerMixin, BaseEstimator):
         self._update_params(X, y)
         self.X_ = X
         self.y_ = y
+        self.efective_min_length = (
+            self.min_length
+            if self.min_length > 1
+            else int(self.min_length * X.shape[0])
+        )
         self.discretizer_ = [None] * self.n_features_in_
         self.cut_points_ = [None] * self.n_features_in_
         Parallel(n_jobs=self.n_jobs, prefer="threads")(
@@ -109,7 +114,7 @@ class FImdlp(TransformerMixin, BaseEstimator):
     def _fit_discretizer(self, feature):
         if feature in self.features_:
             self.discretizer_[feature] = CFImdlp(
-                min_length=self.min_length,
+                min_length=self.efective_min_length,
                 max_depth=self.max_depth,
                 max_cuts=self.max_cuts,
             )