feat: Update DDM to work for classification & regression (#26)

* feat: Implement DDM to work for classification & regression * feat: Implement DDM to work for classification & regression
aaivu · Mar 9, 2024 · b37d028 · b37d028
1 parent 42e27ea
commit b37d028
Show file tree

Hide file tree

Showing 54 changed files with 290 additions and 426 deletions.
diff --git a/src/models/MME4BAT/.gitignore → MME4BAT/.gitignore b/src/models/MME4BAT/.gitignore → MME4BAT/.gitignore
diff --git a/src/models/MME4BAT/LICENSE → MME4BAT/LICENSE b/src/models/MME4BAT/LICENSE → MME4BAT/LICENSE
diff --git a/src/models/MME4BAT/README.md → MME4BAT/README.md b/src/models/MME4BAT/README.md → MME4BAT/README.md
diff --git a/src/models/MME4BAT/convlstm.py → MME4BAT/convlstm.py b/src/models/MME4BAT/convlstm.py → MME4BAT/convlstm.py
diff --git a/src/models/MME4BAT/docs/README.md → MME4BAT/docs/README.md b/src/models/MME4BAT/docs/README.md → MME4BAT/docs/README.md
diff --git a/...odels/MME4BAT/dwelltime_classification.py → MME4BAT/dwelltime_classification.py b/...odels/MME4BAT/dwelltime_classification.py → MME4BAT/dwelltime_classification.py
diff --git a/src/models/MME4BAT/error_analysis.py → MME4BAT/error_analysis.py b/src/models/MME4BAT/error_analysis.py → MME4BAT/error_analysis.py
diff --git a/src/models/MME4BAT/feature_engineering.py → MME4BAT/feature_engineering.py b/src/models/MME4BAT/feature_engineering.py → MME4BAT/feature_engineering.py
diff --git a/...AT/hyperparameter_tuning_and_comparing.py → ...AT/hyperparameter_tuning_and_comparing.py b/...AT/hyperparameter_tuning_and_comparing.py → ...AT/hyperparameter_tuning_and_comparing.py
diff --git a/src/models/MME4BAT/multimodel.py → MME4BAT/multimodel.py b/src/models/MME4BAT/multimodel.py → MME4BAT/multimodel.py
diff --git a/src/models/MME4BAT/outlier_capping.py → MME4BAT/outlier_capping.py b/src/models/MME4BAT/outlier_capping.py → MME4BAT/outlier_capping.py
diff --git a/...efactoring_code/01-feature_engineering.py → ...efactoring_code/01-feature_engineering.py b/...efactoring_code/01-feature_engineering.py → ...efactoring_code/01-feature_engineering.py
diff --git a/...AT/refactoring_code/02-outlier_capping.py → ...AT/refactoring_code/02-outlier_capping.py b/...AT/refactoring_code/02-outlier_capping.py → ...AT/refactoring_code/02-outlier_capping.py
diff --git a/...oring_code/03-dwelltime_classification.py → ...oring_code/03-dwelltime_classification.py b/...oring_code/03-dwelltime_classification.py → ...oring_code/03-dwelltime_classification.py
diff --git a/...s/MME4BAT/refactoring_code/Untitled.ipynb → MME4BAT/refactoring_code/Untitled.ipynb b/...s/MME4BAT/refactoring_code/Untitled.ipynb → MME4BAT/refactoring_code/Untitled.ipynb
diff --git a/...dels/MME4BAT/refactoring_code/convlstm.py → MME4BAT/refactoring_code/convlstm.py b/...dels/MME4BAT/refactoring_code/convlstm.py → MME4BAT/refactoring_code/convlstm.py
diff --git a/...ctoring_code/data/bus_dwell_times_654.csv → ...ctoring_code/data/bus_dwell_times_654.csv b/...ctoring_code/data/bus_dwell_times_654.csv → ...ctoring_code/data/bus_dwell_times_654.csv
diff --git a/...oring_code/data/bus_running_times_654.csv → ...oring_code/data/bus_running_times_654.csv b/...oring_code/data/bus_running_times_654.csv → ...oring_code/data/bus_running_times_654.csv
diff --git a/...data/bus_stop_times_feature_added_all.csv → ...data/bus_stop_times_feature_added_all.csv b/...data/bus_stop_times_feature_added_all.csv → ...data/bus_stop_times_feature_added_all.csv
diff --git a/...data/bus_stop_times_feature_added_new.csv → ...data/bus_stop_times_feature_added_new.csv b/...data/bus_stop_times_feature_added_new.csv → ...data/bus_stop_times_feature_added_new.csv
diff --git a/...code/data/bus_stops_and_terminals_654.csv → ...code/data/bus_stops_and_terminals_654.csv b/...code/data/bus_stops_and_terminals_654.csv → ...code/data/bus_stops_and_terminals_654.csv
diff --git a/...T/refactoring_code/data/bus_trips_654.csv → ...T/refactoring_code/data/bus_trips_654.csv b/...T/refactoring_code/data/bus_trips_654.csv → ...T/refactoring_code/data/bus_trips_654.csv
diff --git a/src/README.md → MME4BAT/src/README.md b/src/README.md → MME4BAT/src/README.md
diff --git a/examples/pipeline.py b/examples/pipeline.py
@@ -9,18 +9,18 @@ def datetime_from_string(datetime_string: str) -> datetime:
 
 if __name__ == "__main__":
     historical_data_starting_from = datetime_from_string("2021-10-01")
-    historical_data_ending_at = datetime_from_string("2021-10-10")
-    streaming_data_starting_from = datetime_from_string("2021-10-10")
-    streaming_data_ending_at = datetime_from_string("2021-11-01")
+    historical_data_ending_at = datetime_from_string("2022-01-01")
+    streaming_data_starting_from = datetime_from_string("2022-01-01")
+    streaming_data_ending_at = datetime_from_string("2022-08-01")
     time_interval = 60
     active_strategy = True
     cdd_strategy = DDM(
-        warning_level=2,
+        warning_level=0.1,
         drift_level=3,
-        min_num_instances=5,
+        min_num_instances=2,
     )
     folder_path_to_save_result = "../experiments"
-    experiment_label = "m-xgb-s-xgb_model"
+    experiment_label = "m-xgb-s-xgb_model-d-batch_r2"
 
     run_exp(
         hist_start=historical_data_starting_from,

diff --git a/src/__init__.py b/src/__init__.py
@@ -1,4 +1,5 @@
-from . import (  # noqa F401
+from . import (
+    concept_drift_detector,
     datasets,
     log_handler,
     models,

diff --git a/src/concept_drift_detector/concept_drift_detector.py b/src/concept_drift_detector/concept_drift_detector.py
@@ -1,11 +1,18 @@
+from pandas import DataFrame
 from src.concept_drift_detector.strategies.istrategy import IStrategy
+# from src.models.base_models import BaseModel
 
 
 class CDD:
     def __init__(self, strategy: IStrategy) -> None:
         self._strategy = strategy
 
-    def is_concept_drift_detected(self, model, ni_x, ni_y) -> bool:
+    def is_concept_drift_detected(
+        self,
+        model,
+        ni_x: DataFrame,
+        ni_y: DataFrame,
+    ) -> bool:
         return self.strategy.is_concept_drift_detected(model, ni_x, ni_y)
 
     @property

diff --git a/src/concept_drift_detector/input.csv b/src/concept_drift_detector/input.csv