wavetrainer 0.0.25__tar.gz → 0.0.26__tar.gz

{wavetrainer-0.0.25/wavetrainer.egg-info → wavetrainer-0.0.26}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: wavetrainer
-Version: 0.0.25
+Version: 0.0.26
 Summary: A library for automatically finding the optimal model within feature and hyperparameter space.
 Home-page: https://github.com/8W9aG/wavetrainer
 Author: Will Sackfield

{wavetrainer-0.0.25 → wavetrainer-0.0.26}/setup.py

@@ -23,7 +23,7 @@ def install_requires() -> typing.List[str]:
 
 setup(
     name='wavetrainer',
-    version='0.0.25',
+    version='0.0.26',
     description='A library for automatically finding the optimal model within feature and hyperparameter space.',
     long_description=long_description,
     long_description_content_type='text/markdown',

{wavetrainer-0.0.25 → wavetrainer-0.0.26}/wavetrainer/__init__.py

@@ -2,5 +2,5 @@
 
 from .create import create
 
-__VERSION__ = "0.0.25"
+__VERSION__ = "0.0.26"
 __all__ = ("create",)

{wavetrainer-0.0.25 → wavetrainer-0.0.26}/wavetrainer/create.py

@@ -15,6 +15,7 @@ def create(
     dt_column: str | None = None,
     max_train_timeout: datetime.timedelta | None = None,
     cutoff_dt: datetime.datetime | None = None,
+    max_features: int | None = None,
 ) -> Trainer:
     """Create a trainer."""
     return Trainer(
@@ -25,4 +26,5 @@ def create(
         dt_column=dt_column,
         max_train_timeout=max_train_timeout,
         cutoff_dt=cutoff_dt,
+        max_features=max_features,
     )

{wavetrainer-0.0.25 → wavetrainer-0.0.26}/wavetrainer/reducer/combined_reducer.py

@@ -25,15 +25,16 @@ class CombinedReducer(Reducer):
 
     # pylint: disable=too-many-positional-arguments,too-many-arguments
 
-    def __init__(self):
+    def __init__(self, max_features: int | None):
         super().__init__()
+        self._max_features = max_features
         self._reducers = [
             UnseenReducer(),
             NonNumericReducer(),
             ConstantReducer(),
             DuplicateReducer(),
             CorrelationReducer(),
-            PCAReducer(),
+            PCAReducer(max_features),
         ]
 
     @classmethod
@@ -62,7 +63,7 @@ class CombinedReducer(Reducer):
                 elif reducer_name == UnseenReducer.name():
                     self._reducers.append(UnseenReducer())
                 elif reducer_name == PCAReducer.name():
-                    self._reducers.append(PCAReducer())
+                    self._reducers.append(PCAReducer(self._max_features))
         for reducer in self._reducers:
             reducer.load(folder)
 

{wavetrainer-0.0.25 → wavetrainer-0.0.26}/wavetrainer/reducer/pca_reducer.py

@@ -20,10 +20,15 @@ class PCAReducer(Reducer):
 
     # pylint: disable=too-many-positional-arguments,too-many-arguments
 
-    def __init__(self):
+    def __init__(self, max_features: int | None):
         super().__init__()
-        self._scaler = StandardScaler()
-        self._pca = PCA(n_components=300)
+        self._max_features = max_features
+        if max_features is not None:
+            self._scaler = StandardScaler()
+            self._pca = PCA(n_components=max_features)
+        else:
+            self._scaler = None
+            self._pca = None
 
     @classmethod
     def name(cls) -> str:
@@ -33,12 +38,18 @@ class PCAReducer(Reducer):
         pass
 
     def load(self, folder: str) -> None:
-        self._scaler = joblib.load(os.path.join(folder, _PCA_SCALER_FILE))
-        self._pca = joblib.load(os.path.join(folder, _PCA_FILE))
+        pca_scaler_file = os.path.join(folder, _PCA_SCALER_FILE)
+        pca_file = os.path.join(folder, _PCA_FILE)
+        if os.path.exists(pca_scaler_file):
+            self._scaler = joblib.load(pca_scaler_file)
+        if os.path.exists(pca_file):
+            self._pca = joblib.load(pca_file)
 
     def save(self, folder: str) -> None:
-        joblib.dump(self._scaler, os.path.join(folder, _PCA_SCALER_FILE))
-        joblib.dump(self._pca, os.path.join(folder, _PCA_FILE))
+        if self._scaler is not None:
+            joblib.dump(self._scaler, os.path.join(folder, _PCA_SCALER_FILE))
+        if self._pca is not None:
+            joblib.dump(self._pca, os.path.join(folder, _PCA_FILE))
 
     def fit(
         self,
@@ -48,13 +59,19 @@ class PCAReducer(Reducer):
         eval_x: pd.DataFrame | None = None,
         eval_y: pd.Series | pd.DataFrame | None = None,
     ) -> Self:
-        if len(df.columns.values) < self._pca.n_components:  # type: ignore
+        pca = self._pca
+        scaler = self._scaler
+        if pca is None or scaler is None:
+            return self
+        if len(df.columns.values) < pca.n_components:  # type: ignore
             return self
-        x_scaled = self._scaler.fit_transform(df)
-        self._pca.fit(x_scaled)
+        x_scaled = scaler.fit_transform(df)
+        pca.fit(x_scaled)
         return self
 
     def transform(self, df: pd.DataFrame) -> pd.DataFrame:
+        if self._pca is None:
+            return df
         if len(df.columns.values) < self._pca.n_components:  # type: ignore
             return df
         return self._pca.transform(df)

{wavetrainer-0.0.25 → wavetrainer-0.0.26}/wavetrainer/trainer.py

@@ -36,6 +36,7 @@ _TEST_SIZE_KEY = "test_size"
 _VALIDATION_SIZE_KEY = "validation_size"
 _IDX_USR_ATTR_KEY = "idx"
 _DT_COLUMN_KEY = "dt_column"
+_MAX_FEATURES_KEY = "max_features"
 
 
 class Trainer(Fit):
@@ -53,6 +54,7 @@ class Trainer(Fit):
         dt_column: str | None = None,
         max_train_timeout: datetime.timedelta | None = None,
         cutoff_dt: datetime.datetime | None = None,
+        max_features: int | None = None,
     ):
         tqdm.tqdm.pandas()
 
@@ -103,6 +105,7 @@ class Trainer(Fit):
                         )
                 if dt_column is None:
                     dt_column = params[_DT_COLUMN_KEY]
+                max_features = params.get(_MAX_FEATURES_KEY)
         else:
             with open(params_file, "w", encoding="utf8") as handle:
                 validation_size_value = None
@@ -133,6 +136,7 @@ class Trainer(Fit):
                         _TEST_SIZE_KEY: test_size_value,
                         _VALIDATION_SIZE_KEY: validation_size_value,
                         _DT_COLUMN_KEY: dt_column,
+                        _MAX_FEATURES_KEY: max_features,
                     },
                     handle,
                 )
@@ -143,6 +147,7 @@ class Trainer(Fit):
         self._dt_column = dt_column
         self._max_train_timeout = max_train_timeout
         self._cutoff_dt = cutoff_dt
+        self._max_features = max_features
 
     def _provide_study(self, column: str) -> optuna.Study:
         storage_name = f"sqlite:///{self._folder}/{column}/{_STUDYDB_FILENAME}"
@@ -216,7 +221,7 @@ class Trainer(Fit):
                         return -1.0
 
                     # Perform common reductions
-                    reducer = CombinedReducer()
+                    reducer = CombinedReducer(self._max_features)
                     reducer.set_options(trial)
                     x_train = reducer.fit_transform(x_train)
                     x_test = reducer.transform(x_test)
@@ -425,7 +430,7 @@ class Trainer(Fit):
                 date_str = dates[-1].isoformat()
                 folder = os.path.join(column_path, date_str)
 
-                reducer = CombinedReducer()
+                reducer = CombinedReducer(self._max_features)
                 reducer.load(folder)
 
                 model = ModelRouter()

{wavetrainer-0.0.25 → wavetrainer-0.0.26/wavetrainer.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: wavetrainer
-Version: 0.0.25
+Version: 0.0.26
 Summary: A library for automatically finding the optimal model within feature and hyperparameter space.
 Home-page: https://github.com/8W9aG/wavetrainer
 Author: Will Sackfield