make sure scores_df is always freshly created

autoemulate/compare.py

@@ -74,11 +74,6 @@ def setup(
         self.n_jobs = n_jobs
         self.logger = configure_logging(log_to_file=log_to_file)
         self.is_set_up = True
-        self.scores_df = pd.DataFrame(
-            columns=["model", "metric", "fold", "score"]
-        ).astype(
-            {"model": "object", "metric": "object", "fold": "int64", "score": "float64"}
-        )
         self.cv_results = {}
 
     def compare(self):
@@ -92,6 +87,13 @@ def compare(self):
         if not self.is_set_up:
             raise RuntimeError("Must run setup() before compare()")
 
+        # Freshly initialise scores dataframe when running compare()
+        self.scores_df = pd.DataFrame(
+            columns=["model", "metric", "fold", "score"]
+        ).astype(
+            {"model": "object", "metric": "object", "fold": "int64", "score": "float64"}
+        )
+
         for model in self.models:
             # search for best hyperparameters
             if self.hyperparameter_search:
@@ -136,7 +138,7 @@ def cross_validate(self, model):
         self.logger.info(f"Parameters: {model.get_params()}")
 
         # Cross-validate
-        cv = cross_validate(
+        cv_results = cross_validate(
             model,
             self.X,
             self.y,
@@ -146,20 +148,37 @@ def cross_validate(self, model):
             return_estimator=True,
             return_indices=True,
         )
+        self._update_scores_df(model_name, cv_results)
 
+    def _update_scores_df(self, model_name, cv_results):
+        """Updates the scores dataframe with the results of the cross-validation.
+
+        Parameters
+        ----------
+            model_name : str
+                Name of the model.
+            cv_results : dict
+                Results of the cross-validation.
+
+        Returns
+        -------
+            scores_df : pandas.DataFrame
+                Dataframe containing the scores for each model, metric and fold.
+
+        """
         # Gather scores from each metric
         # Initialise scores dataframe
-        for key in cv.keys():
+        for key in cv_results.keys():
             if key.startswith("test_"):
-                for fold, score in enumerate(cv[key]):
+                for fold, score in enumerate(cv_results[key]):
                     self.scores_df.loc[len(self.scores_df.index)] = {
                         "model": model_name,
                         "metric": key.split("test_", 1)[1],
                         "fold": fold,
                         "score": score,
                     }
         # save results for plotting etc.
-        self.cv_results[model_name] = cv
+        self.cv_results[model_name] = cv_results
 
     def hyperparam_search(self, model):
         """Performs hyperparameter search for a given model.