Created Tree_Parzen_Estimator function

Author: LucaQuade

lkauto/optimization_strategies/tree_parzen_estimator.py

@@ -0,0 +1,35 @@
+import numpy as np
+import pandas as pd
+import time
+
+from ConfigSpace import ConfigurationSpace, Configuration
+
+from lenskit.data import Dataset, ItemListCollection
+
+from lkauto.explicit.explicit_evaler import ExplicitEvaler
+from lkauto.implicit.implicit_evaler import ImplicitEvaler
+from lkauto.utils.get_default_configurations import get_default_configurations
+from lkauto.utils.filer import Filer
+from lkauto.utils.get_default_configuration_space import get_default_configuration_space
+
+from hyperopt import fmin, tpe, space_eval
+
+from typing import Tuple
+import logging
+
+
+def tree_parzen(cs: ConfigurationSpace,
+                  train: Dataset,
+                  user_feedback: str,
+                  optimization_metric,
+                  filer: Filer,
+                  validation: ItemListCollection = None,
+                  time_limit_in_sec: int = 3600,
+                  num_evaluations: int = None,
+                  split_folds: int = 1,
+                  split_strategie: str = 'user_based',
+                  split_frac: float = 0.25,
+                  ensemble_size: int = 50,
+                  minimize_error_metric_val: bool = True,
+                  num_recommendations: int = 10,
+                  random_state=42) -> Tuple[Configuration, pd.DataFrame]: