34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453 | class FeatureEngineer:
"""
Orchestrates a sequence of feature engineering steps.
"""
def __init__(self, steps_config: List[Dict[str, Any]]):
self.steps_config = steps_config
self.fitted_steps: List[Dict[str, Any]] = []
def transform(self, data: pd.DataFrame) -> pd.DataFrame:
"""
Apply fitted transformations to new data.
"""
current_data = data
for step in self.fitted_steps:
name = step["name"]
transformer_type = step["type"]
applier = step["applier"]
artifact = step["artifact"]
# Skip splitters during inference/transform
if transformer_type in [
"TrainTestSplitter",
"feature_target_split",
"Oversampling",
"Undersampling",
]:
continue
logger.debug(f"Applying step: {name} ({transformer_type})")
current_data = applier.apply(current_data, artifact)
return current_data
def fit_transform(self, data: Union[pd.DataFrame, Any], node_id_prefix="") -> Any: # noqa: C901
"""
Runs the pipeline on data.
Returns: (transformed_data, metrics_dict)
"""
self.fitted_steps = [] # Reset fitted steps
current_data = data
metrics: Dict[str, Any] = {}
for i, step in enumerate(self.steps_config):
name = step["name"]
transformer_type = step["transformer"]
params = step.get("params", {})
logger.info(f"Running step {i}: {name} ({transformer_type})")
logger.debug(
f"FeatureEngineer running step {i}: {name} ({transformer_type})"
)
logger.debug(f"current_data type: {type(current_data)}")
# Capture metrics before
rows_before, cols_before = get_data_stats(current_data)
# Keep reference for comparison (for Winsorize metrics)
data_before = current_data
calculator, applier = self._get_transformer_components(transformer_type)
# We need a unique ID for this step's artifacts
step_node_id = f"{node_id_prefix}_{name}"
transformer = StatefulTransformer(calculator, applier, step_node_id)
# Handle special transformers that change data structure
# Splitters return SplitDataset or (X, y) tuples instead of a simple DataFrame,
# so they bypass the standard StatefulTransformer wrapper.
# Initialize fitted_params
fitted_params = {}
if transformer_type == "TrainTestSplitter":
logger.debug("Handling TrainTestSplitter")
# TrainTestSplitter changes DataFrame -> SplitDataset.
# We bypass StatefulTransformer to allow this structural change.
# It can also handle (X, y) tuple if FeatureTargetSplit was done first.
if isinstance(current_data, (pd.DataFrame, tuple)):
logger.debug("Executing TrainTestSplitter logic")
params = calculator.fit(current_data, params)
current_data = applier.apply(current_data, params)
# In SDK, params are returned but not auto-saved to artifact store here.
# The Pipeline object will handle state persistence.
else:
logger.debug(
f"Skipping TrainTestSplitter. current_data is {type(current_data)}"
)
logger.warning(
"Attempting to split an already split dataset. Skipping TrainTestSplitter."
)
elif transformer_type == "feature_target_split":
logger.debug("Handling feature_target_split")
# FeatureTargetSplitter changes structure to (X, y) or Dict of (X, y).
# We bypass StatefulTransformer to allow this structural change.
params = calculator.fit(current_data, params)
current_data = applier.apply(current_data, params)
else:
logger.debug("Handling standard transformer via StatefulTransformer")
current_data = transformer.fit_transform(current_data, params)
# In SDK, transformer.params holds the state.
fitted_params = transformer.params
self.fitted_steps.append(
{
"name": name,
"type": transformer_type,
"applier": applier,
"artifact": fitted_params,
}
)
logger.debug(f"Step {i} complete. New data type: {type(current_data)}")
# Retrieve fitted params to get metrics from the calculator
try:
if fitted_params:
# Imputation Metrics
if transformer_type in [
"SimpleImputer",
"KNNImputer",
"IterativeImputer",
]:
if "missing_counts" in fitted_params:
metrics["missing_counts"] = fitted_params["missing_counts"]
if "total_missing" in fitted_params:
metrics["total_missing"] = fitted_params["total_missing"]
if "fill_values" in fitted_params:
metrics["fill_values"] = fitted_params["fill_values"]
# Feature Selection Metrics
if transformer_type in [
"feature_selection",
"UnivariateSelection",
"ModelBasedSelection",
"VarianceThreshold",
]:
if "feature_scores" in fitted_params:
metrics["feature_scores"] = fitted_params["feature_scores"]
if "p_values" in fitted_params:
metrics["p_values"] = fitted_params["p_values"]
if "feature_importances" in fitted_params:
metrics["feature_importances"] = fitted_params[
"feature_importances"
]
if "variances" in fitted_params:
metrics["variances"] = fitted_params["variances"]
if "ranking" in fitted_params:
metrics["ranking"] = fitted_params["ranking"]
if "selected_columns" in fitted_params:
metrics["selected_columns"] = fitted_params[
"selected_columns"
]
# Scaling Metrics
if transformer_type in [
"StandardScaler",
"MinMaxScaler",
"RobustScaler",
"MaxAbsScaler",
]:
if "mean" in fitted_params:
metrics["mean"] = fitted_params["mean"]
if "scale" in fitted_params:
metrics["scale"] = fitted_params["scale"]
if "var" in fitted_params:
metrics["var"] = fitted_params["var"]
if "min" in fitted_params:
metrics["min"] = fitted_params["min"]
if "data_min" in fitted_params:
metrics["data_min"] = fitted_params["data_min"]
if "data_max" in fitted_params:
metrics["data_max"] = fitted_params["data_max"]
if "center" in fitted_params:
metrics["center"] = fitted_params["center"]
if "max_abs" in fitted_params:
metrics["max_abs"] = fitted_params["max_abs"]
if "columns" in fitted_params:
metrics["columns"] = fitted_params["columns"]
# Outlier Metrics
if transformer_type in [
"IQR",
"Winsorize",
"ZScore",
"EllipticEnvelope",
]:
if "warnings" in fitted_params:
metrics["warnings"] = fitted_params["warnings"]
if transformer_type in ["IQR", "Winsorize"]:
if "bounds" in fitted_params:
metrics["bounds"] = fitted_params["bounds"]
if transformer_type == "ZScore":
if "stats" in fitted_params:
metrics["stats"] = fitted_params["stats"]
if transformer_type == "EllipticEnvelope":
if "contamination" in fitted_params:
metrics["contamination"] = fitted_params["contamination"]
# Bucketing Metrics
if transformer_type in [
"GeneralBinning",
"EqualWidthBinning",
"EqualFrequencyBinning",
"CustomBinning",
"KBinsDiscretizer",
]:
if "bin_edges" in fitted_params:
metrics["bin_edges"] = fitted_params["bin_edges"]
if "n_bins" in fitted_params:
metrics["n_bins"] = fitted_params["n_bins"]
# Feature Generation Metrics
if transformer_type in ["FeatureMath", "FeatureGenerationNode"]:
if "operations" in fitted_params:
metrics["operations_count"] = len(
fitted_params["operations"]
)
metrics["operations"] = fitted_params["operations"]
# Calculate generated features by comparing columns
if isinstance(data_before, pd.DataFrame) and isinstance(
current_data, pd.DataFrame
):
new_cols = list(
set(current_data.columns) - set(data_before.columns)
)
metrics["generated_features"] = new_cols
elif isinstance(data_before, SplitDataset) and isinstance(
current_data, SplitDataset
):
# Check train set
if isinstance(
data_before.train, pd.DataFrame
) and isinstance(current_data.train, pd.DataFrame):
new_cols = list(
set(current_data.train.columns)
- set(data_before.train.columns)
)
metrics["generated_features"] = new_cols
elif isinstance(data_before.train, tuple) and isinstance(
current_data.train, tuple
):
# (X, y) tuple
X_before, _ = data_before.train
X_after, _ = current_data.train
if isinstance(X_before, pd.DataFrame) and isinstance(
X_after, pd.DataFrame
):
new_cols = list(
set(X_after.columns) - set(X_before.columns)
)
metrics["generated_features"] = new_cols
except Exception as e:
logger.warning(f"Failed to retrieve metrics for step {name}: {e}")
# Capture metrics after
rows_after, cols_after = get_data_stats(current_data)
# Resampling Metrics (Calculated from data)
if transformer_type in ["Oversampling", "Undersampling"]:
try:
# Extract y to calculate class counts
y_res = None
if isinstance(current_data, SplitDataset):
if isinstance(current_data.train, tuple):
_, y_res = current_data.train
elif isinstance(current_data.train, pd.DataFrame):
# Try to find target column from params
target_col = params.get("target_column")
if target_col and target_col in current_data.train.columns:
y_res = current_data.train[target_col]
elif isinstance(current_data, tuple):
_, y_res = current_data
elif isinstance(current_data, pd.DataFrame):
target_col = params.get("target_column")
if target_col and target_col in current_data.columns:
y_res = current_data[target_col]
if y_res is not None:
counts = y_res.value_counts().to_dict()
# Convert keys to string to ensure JSON serializability
metrics["class_counts"] = {
str(k): int(v) for k, v in counts.items()
}
metrics["total_samples"] = int(len(y_res))
except Exception as e:
logger.warning(f"Failed to calculate resampling metrics: {e}")
if rows_after > 0 or cols_after:
if transformer_type in [
"DropMissingRows",
"Deduplicate",
"IQR",
"ZScore",
"EllipticEnvelope",
"Winsorize",
]:
dropped = rows_before - rows_after
metrics[f"{transformer_type}_rows_removed"] = dropped
metrics[f"{transformer_type}_rows_remaining"] = rows_after
metrics[f"{transformer_type}_rows_total"] = rows_before
metrics["rows_removed"] = dropped
metrics["rows_total"] = rows_before
# Special metric for Winsorize: Values Clipped
if transformer_type == "Winsorize":
try:
clipped_count = 0
# Helper to count diffs
def count_diffs(df1, df2):
if isinstance(df1, pd.DataFrame) and isinstance(
df2, pd.DataFrame
):
if df1.shape == df2.shape:
return int(df1.ne(df2).sum().sum())
elif (
isinstance(df1, tuple)
and isinstance(df2, tuple)
and len(df1) == 2
and len(df2) == 2
):
# Handle (X, y) tuple
diffs = 0
# Compare X (index 0)
if isinstance(df1[0], pd.DataFrame) and isinstance(
df2[0], pd.DataFrame
):
if df1[0].shape == df2[0].shape:
diffs += int(df1[0].ne(df2[0]).sum().sum())
# Compare y (index 1) - usually Series
if isinstance(
df1[1], (pd.DataFrame, pd.Series)
) and isinstance(df2[1], (pd.DataFrame, pd.Series)):
if df1[1].shape == df2[1].shape:
diffs += int(df1[1].ne(df2[1]).sum().sum()) # type: ignore
return diffs
return 0
if isinstance(data_before, pd.DataFrame) and isinstance(
current_data, pd.DataFrame
):
clipped_count = count_diffs(data_before, current_data)
elif isinstance(data_before, SplitDataset) and isinstance(
current_data, SplitDataset
):
clipped_count += count_diffs(
data_before.train, current_data.train
)
clipped_count += count_diffs(
data_before.test, current_data.test
)
clipped_count += count_diffs(
data_before.validation, current_data.validation
)
metrics["values_clipped"] = clipped_count
except Exception as e:
logger.warning(
f"Failed to calculate values_clipped for Winsorize: {e}"
)
pass
if transformer_type == "MissingIndicator":
new_cols_set = cols_after - cols_before
metrics["missing_indicators_created"] = len(new_cols_set)
cast(Dict[str, Any], metrics)["missing_indicators_columns"] = list(
new_cols_set
)
if transformer_type == "DropMissingColumns":
dropped_cols_set = cols_before - cols_after
cast(Dict[str, Any], metrics)["dropped_columns"] = list(
dropped_cols_set
)
metrics["dropped_columns_count"] = len(dropped_cols_set)
if transformer_type == "feature_selection":
dropped_cols_set = cols_before - cols_after
cast(Dict[str, Any], metrics)["dropped_columns"] = list(
dropped_cols_set
)
metrics["dropped_columns_count"] = len(dropped_cols_set)
if transformer_type in [
"OneHotEncoder",
"LabelEncoder",
"OrdinalEncoder",
"TargetEncoder",
"HashEncoder",
"DummyEncoder",
]:
new_cols_set = cols_after - cols_before
metrics["new_features_count"] = len(new_cols_set)
metrics["encoded_columns_count"] = len(params.get("columns", []))
if "categories_count" in params:
metrics["categories_count"] = params["categories_count"]
if "classes_count" in params:
metrics["classes_count"] = params["classes_count"]
return current_data, metrics
def _get_transformer_components(self, type_name: str): # noqa: C901
# Try Registry first
try:
return (
NodeRegistry.get_calculator(type_name)(),
NodeRegistry.get_applier(type_name)(),
)
except ValueError:
raise ValueError(f"Unknown transformer type: {type_name}")
|