Miscellaneous fixes relevant to text2time, time, timeseries AIT modul…

…e tutorials (#100) * fixes associated with text module * fixes associated with AIT tutorial --------- Co-authored-by: kbonney <[email protected]>
sandialabs · Oct 25, 2024 · bc345b5 · bc345b5
1 parent be87631
commit bc345b5
Show file tree

Hide file tree

Showing 6 changed files with 571 additions and 646 deletions.
diff --git a/pvops/text/classify.py b/pvops/text/classify.py
@@ -97,7 +97,7 @@ def classification_deployer(
 
         .. code-block:: python
 
-            scoring = make_scorer(f1_score, average = 'weighted')
+            scoring = make_scorer(f1_score, average = 'weighted', pos_label = None)
 
         For unsupervised classifiers, one could specify
 

diff --git a/pvops/text/defaults.py b/pvops/text/defaults.py
@@ -90,7 +90,7 @@ def supervised_classifier_defs(settings_flag):
                 "clf__C": np.logspace(0, 4, 10),
             },
             "PassiveAggressiveClassifier": {
-                "clf__C": [0.0, 0.01, 0.1, 1.0],
+                "clf__C": [0.001, 0.01, 0.1, 1.0],
                 "clf__loss": ["hinge", "squared_hinge"],
             },
             "RidgeClassifier": {
@@ -119,7 +119,7 @@ def supervised_classifier_defs(settings_flag):
             "AdaBoostClassifier": {
                 "clf__n_estimators": [50, 100],
                 "clf__learning_rate": [1.0, 0.9, 0.8],
-                "clf__algorithm": ["SAMME.R"],
+                "clf__algorithm": ["SAMME"],
             },
         }
     elif settings_flag == "normal":
@@ -152,7 +152,7 @@ def supervised_classifier_defs(settings_flag):
                 "clf__C": np.logspace(0, 4, 10),
             },
             "PassiveAggressiveClassifier": {
-                "clf__C": [0.0, 0.01, 0.1, 1.0],
+                "clf__C": [0.001, 0.01, 0.1, 1.0],
                 "clf__loss": ["hinge", "squared_hinge"],
             },
             "RidgeClassifier": {
@@ -182,7 +182,7 @@ def supervised_classifier_defs(settings_flag):
             "AdaBoostClassifier": {
                 "clf__n_estimators": [50, 100],
                 "clf__learning_rate": [1.0, 0.9, 0.8],
-                "clf__algorithm": ["SAMME.R"],
+                "clf__algorithm": ["SAMME"],
             },
         }
     elif settings_flag == "detailed":
@@ -215,7 +215,7 @@ def supervised_classifier_defs(settings_flag):
                 "clf__C": np.logspace(0, 4, 10),
             },
             "PassiveAggressiveClassifier": {
-                "clf__C": [0.0, 0.01, 0.1, 1.0],
+                "clf__C": [0.001, 0.01, 0.1, 1.0],
                 "clf__loss": ["hinge", "squared_hinge"],
             },
             "RidgeClassifier": {
@@ -245,7 +245,7 @@ def supervised_classifier_defs(settings_flag):
             "AdaBoostClassifier": {
                 "clf__n_estimators": [30, 50, 100, 150, 300],
                 "clf__learning_rate": [1.0, 0.9, 0.8, 0.4],
-                "clf__algorithm": ["SAMME.R", "SAMME"],
+                "clf__algorithm": ["SAMME"],
             },
         }
 

diff --git a/pvops/timeseries/models/AIT.py b/pvops/timeseries/models/AIT.py
@@ -126,12 +126,11 @@ def _load_params(self):
                                    (0.06624491753542901, [0, 0])]
 
     def predict_subset(self, prod_df, scaler, model_terms, prod_col_dict):
-        prod_df = prod_df.copy()
         self.check_data(prod_df, prod_col_dict)
 
         """1. Standardize the data using same scales"""
         for k, d in scaler.items():
-            data = prod_df[prod_col_dict[k]]
+            data = prod_df[prod_col_dict[k]].copy()
             scaled_data = self._apply_transform(data, d)
             prod_df[self._col_scaled_prefix + prod_col_dict[k]] = scaled_data
 

diff --git a/tutorials/text_class_example.py b/tutorials/text_class_example.py
@@ -349,7 +349,7 @@ def classify_supervised(
             pipeline_steps = [
                 ("doc2vec", nlp_utils.Doc2VecModel()), ("clf", None)]
 
-        scoring = make_scorer(f1_score, average="weighted")
+        scoring = make_scorer(f1_score, average="weighted", pos_label=None)
         self.greater_is_better = True
 
         if user_defined_classes is None or user_defined_search_space is None:

diff --git a/tutorials/tutorial_AIT_timeseries.ipynb b/tutorials/tutorial_AIT_timeseries.ipynb
diff --git a/tutorials/tutorial_textmodule.ipynb b/tutorials/tutorial_textmodule.ipynb