рефакторинг блокнота research в части логирования в mlflow

iis_project/sklearn_utils/__init__.py

@@ -0,0 +1,60 @@
+from collections.abc import Container, Sequence, Mapping
+from typing import TypeAlias, TypeVar
+
+
+ParamsFilterSpec: TypeAlias = (
+    bool
+    | Container[str]
+    | tuple[bool, Container[str]]
+    | Mapping[str, 'ParamsFilterSpec']
+    | tuple[bool, 'ParamsFilterSpec']
+)
+
+
+V = TypeVar('V')
+
+
+def _split_param_key(key: str) -> tuple[str, ...]:
+    return tuple(key.split('__'))
+
+
+def _match_key_to_filter_spec(
+    key: Sequence[str], spec: ParamsFilterSpec, empty_default: bool,
+) -> bool:
+    if isinstance(spec, Sequence) and (len(spec) == 2) and isinstance(spec[0], bool):
+        if (len(key) == 0) and (not spec[0]):
+            return empty_default
+        spec = spec[1]
+    if isinstance(spec, Mapping):
+        if len(key) == 0:
+            return empty_default
+        spec_nested = spec.get(key[0])
+        if spec_nested is None:
+            return False
+        return _whether_to_include_param(key[1:], spec_nested)
+    elif isinstance(spec, Container):
+        if len(key) == 0:
+            return True
+        return (key[0] in spec)
+    return bool(spec)
+
+
+def _whether_to_include_param(
+    key: Sequence[str], include: ParamsFilterSpec = True, exclude: ParamsFilterSpec = False,
+) -> bool:
+    return (
+        (not _match_key_to_filter_spec(key, exclude, empty_default=False))
+        and _match_key_to_filter_spec(key, include, empty_default=True)
+    )
+
+
+def filter_params(
+    params: Mapping[str, V],
+    include: ParamsFilterSpec = True,
+    exclude: ParamsFilterSpec = False,
+) -> Mapping[str, V]:
+    return {
+        k: v
+        for k, v in params.items()
+        if _whether_to_include_param(_split_param_key(k), include, exclude)
+    }

iis_project/sklearn_utils/compose.py

@@ -0,0 +1,3 @@
+COLUMN_TRANSFORMER_PARAMS_COMMON_INCLUDE = [
+    'remainder', 'sparse_threshold', 'transformer_weights',
+]

iis_project/sklearn_utils/ensemble.py

@@ -0,0 +1 @@
+RANDOM_FOREST_REGRESSOR_PARAMS_COMMON_EXCLUDE = ['n_jobs', 'verbose', 'warm_start']

iis_project/sklearn_utils/pipeline.py

@@ -0,0 +1 @@
+PIPELINE_PARAMS_COMMON_INCLUDE = ['transform_input']

iis_project/sklearn_utils/preprocessing.py

@@ -0,0 +1 @@
+STANDARD_SCALER_PARAMS_COMMON_EXCLUDE = ['copy']

research/research.py

@@ -51,6 +51,7 @@ mlflow_run_name: str = 'Baseline model'
 import os
 import pathlib
 import pickle
+import sys
 
 # %%
 import mlflow
@@ -66,6 +67,17 @@ import sklearn.preprocessing
 # %%
 BASE_PATH = pathlib.Path('..')
 
+# %%
+CODE_PATH = BASE_PATH
+sys.path.insert(0, str(CODE_PATH.resolve()))
+
+# %%
+from iis_project.sklearn_utils import filter_params
+from iis_project.sklearn_utils.compose import COLUMN_TRANSFORMER_PARAMS_COMMON_INCLUDE
+from iis_project.sklearn_utils.ensemble import RANDOM_FOREST_REGRESSOR_PARAMS_COMMON_EXCLUDE
+from iis_project.sklearn_utils.pipeline import PIPELINE_PARAMS_COMMON_INCLUDE
+from iis_project.sklearn_utils.preprocessing import STANDARD_SCALER_PARAMS_COMMON_EXCLUDE
+
 # %%
 MODEL_INOUT_EXAMPLE_SIZE = 0x10
 
@@ -196,6 +208,22 @@ tuple(map(len, (df_target_train, df_target_test)))
 mlflow_model_signature = mlflow.models.infer_signature(model_input=df_orig_features, model_output=df_target)
 mlflow_model_signature
 
+# %% [raw] vscode={"languageId": "raw"}
+# input_schema = mlflow.types.schema.Schema([
+#     mlflow.types.schema.ColSpec("double", "selling_price"),
+#     mlflow.types.schema.ColSpec("double", "driven_kms"),
+#     mlflow.types.schema.ColSpec("string", "fuel_type"),
+#     mlflow.types.schema.ColSpec("string", "selling_type"),
+#     mlflow.types.schema.ColSpec("string", "transmission"),
+#     mlflow.types.schema.ColSpec("double", "age"),
+# ])
+#
+# output_schema = mlflow.types.schema.Schema([
+#     mlflow.types.schema.ColSpec("double", "present_price"),
+# ])
+#
+# mlflow_model_signature = mlflow.models.ModelSignature(inputs=input_schema, outputs=output_schema)
+
 # %% [markdown]
 # Пайплайн предобработки признаков:
 
@@ -235,7 +263,25 @@ pipeline = sklearn.pipeline.Pipeline([
 pipeline
 
 # %%
-model_params = pipeline.get_params()
+model_params = filter_params(
+    pipeline.get_params(),
+    include={
+        **{k: True for k in PIPELINE_PARAMS_COMMON_INCLUDE},
+        'preprocess': (
+            False,
+            {
+                **{k: True for k in COLUMN_TRANSFORMER_PARAMS_COMMON_INCLUDE},
+                'scale_to_standard': True,
+                'encode_categorical_wrt_target': True,
+            },
+        ),
+        'regress': (False, True),
+    },
+    exclude={
+        'preprocess': {'scale_to_standard': STANDARD_SCALER_PARAMS_COMMON_EXCLUDE},
+        'regress': RANDOM_FOREST_REGRESSOR_PARAMS_COMMON_EXCLUDE,
+    },
+)
 model_params
 
 # %% [markdown]