Merge pull request #51 from alan-turing-institute/fix-hyperparam-search

Fix hyperparam search

autoemulate/compare.py

@@ -142,7 +142,7 @@ def compare(self):
 
         for i, model in enumerate(self.models):
             updated_model = (
-                self._get_best_hyperparams_hyperparams(i, model)
+                self._get_best_hyperparams(i, model)
                 if self.hyperparameter_search
                 else model
             )
@@ -243,7 +243,12 @@ def _get_best_hyperparams(self, model_index, model):
         """
         # Perform hyperparameter search and update model
         hyperparam_searcher = HyperparamSearch(
-            self.X, self.y, self.cv, self.n_jobs, self.logger
+            X=self.X,
+            y=self.y,
+            cv=self.cv,
+            param_grid=None,
+            n_jobs=self.n_jobs,
+            logger=self.logger,
         )
         updated_model = hyperparam_searcher.search(model)
 
@@ -299,7 +304,7 @@ def _get_best_model(self, metric="r2"):
 
         return best_model
 
-    def print_scores(self, model=None):
+    def print_results(self, model=None):
         # check if model is in self.models
         if model is not None:
             model_names = [type(model).__name__ for model in self.models]

autoemulate/datasets.py

@@ -20,7 +20,8 @@ def fetch_dataset(dataset_name, train_test=False, test_size=0.2, random_state=42
         "four_chamber" is the 4-chamber model without LH sampling.
         "circ_adapt" is the circulatory adaptation model also from LH sampling.
     train_test : bool, optional
-        If True, returns the dataset split into training and testing sets.
+        If True, returns the dataset split into training and testing sets,
+        X_train, X_test, y_train, y_test.
         If False, returns the entire dataset. Default is False.
     test_size : float, optional
         The proportion of the dataset to include in the test split.

autoemulate/hyperparam_search.py

@@ -33,7 +33,7 @@ def __init__(self, X, y, cv, n_jobs, param_grid=None, logger=None):
         self.cv = cv
         self.n_jobs = n_jobs
         self.param_grid = param_grid
-        self.logger = logger or logging.getLogger(__name__)
+        self.logger = logger
         self.best_params = {}
 
     def search(self, model):
@@ -64,7 +64,7 @@ def prepare_param_grid(model, param_grid=None):
         """Prepares the parameter grid with prefixed parameters."""
         if param_grid is None:
             param_grid = model.named_steps["model"].get_grid_params()
-        print(f"param_grid: {param_grid}")
+        # print(f"param_grid: {param_grid}")
         return {f"model__{key}": value for key, value in param_grid.items()}
 
     @staticmethod