Spaces:
No application file
No application file
File size: 17,906 Bytes
669d6a1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 | """
Specialized caching for cross-validation functions.
Handles sklearn classifiers, CV generators, and complex ML workflows.
"""
import hashlib
import inspect
import json
import pickle
import time
from functools import wraps
from typing import Callable, Optional
import numpy as np
import pandas as pd
from loguru import logger
from sklearn.base import BaseEstimator
def _hash_classifier(clf: BaseEstimator) -> str:
"""
Generate stable hash for sklearn classifier.
Uses class name + parameters (not the trained state).
"""
try:
# Get classifier type and parameters
clf_type = type(clf).__name__
params = clf.get_params(deep=True)
# Filter out non-serializable params (like objects, functions)
serializable_params = {}
for k, v in params.items():
try:
# Test if JSON serializable
json.dumps(v)
serializable_params[k] = v
except (TypeError, ValueError):
# Use type name for non-serializable params
serializable_params[k] = f"<{type(v).__name__}>"
# Create stable hash
param_str = json.dumps(serializable_params, sort_keys=True)
combined = f"{clf_type}_{param_str}"
return hashlib.md5(combined.encode()).hexdigest()[:12]
except Exception as e:
logger.debug(f"Failed to hash classifier: {e}")
return f"clf_{type(clf).__name__}_{id(clf)}"
def _hash_cv_generator(cv_gen) -> str:
"""Generate hash for CV generator (KFold, PurgedKFold, etc.)"""
try:
cv_type = type(cv_gen).__name__
# Get CV parameters
params = {}
if hasattr(cv_gen, "n_splits"):
params["n_splits"] = cv_gen.n_splits
if hasattr(cv_gen, "pct_embargo"):
params["pct_embargo"] = cv_gen.pct_embargo
if hasattr(cv_gen, "t1"):
# Hash the t1 series structure (not full content for speed)
t1 = cv_gen.t1
if isinstance(t1, pd.Series):
params["t1_len"] = len(t1)
params["t1_start"] = str(t1.index[0])
params["t1_end"] = str(t1.index[-1])
param_str = json.dumps(params, sort_keys=True)
combined = f"{cv_type}_{param_str}"
return hashlib.md5(combined.encode()).hexdigest()[:12]
except Exception as e:
logger.debug(f"Failed to hash CV generator: {e}")
return f"cv_{type(cv_gen).__name__}_{id(cv_gen)}"
def _hash_dataframe_fast(df: pd.DataFrame) -> str:
"""
Fast DataFrame hashing for CV caching.
Uses shape + columns + index range + sample of data.
"""
parts = [
f"shape_{df.shape}",
f"cols_{hashlib.md5(str(tuple(df.columns)).encode()).hexdigest()[:8]}",
]
# Hash index
if isinstance(df.index, pd.DatetimeIndex):
parts.append(f"idx_{df.index[0]}_{df.index[-1]}_{len(df)}")
else:
parts.append(f"idx_{df.index[0]}_{df.index[-1]}")
# Sample data for hash (for speed)
if len(df) > 100:
sample = df.iloc[:: max(1, len(df) // 100)]
else:
sample = df
data_hash = hashlib.md5(sample.values.tobytes()).hexdigest()[:8]
parts.append(f"data_{data_hash}")
return "_".join(parts)
def _hash_series_fast(series: pd.Series) -> str:
"""Fast Series hashing."""
parts = [f"len_{len(series)}", f"dtype_{series.dtype}"]
if isinstance(series.index, pd.DatetimeIndex):
parts.append(f"idx_{series.index[0]}_{series.index[-1]}")
# Sample for hash
if len(series) > 100:
sample = series.iloc[:: max(1, len(series) // 100)]
else:
sample = series
data_hash = hashlib.md5(sample.values.tobytes()).hexdigest()[:8]
parts.append(f"data_{data_hash}")
return "_".join(parts)
def _generate_cv_cache_key(func: Callable, args: tuple, kwargs: dict) -> str:
"""
Generate specialized cache key for CV functions.
Handles classifiers, CV generators, DataFrames, and sample weights.
"""
key_parts = [func.__module__, func.__qualname__]
# Get function signature to map args to param names
sig = inspect.signature(func)
bound = sig.bind(*args, **kwargs)
bound.apply_defaults()
for param_name, param_value in bound.arguments.items():
try:
# Handle different parameter types
if param_value is None:
key_parts.append(f"{param_name}_None")
elif isinstance(param_value, BaseEstimator):
# Sklearn classifier/estimator
clf_hash = _hash_classifier(param_value)
key_parts.append(f"{param_name}_clf_{clf_hash}")
elif hasattr(param_value, "split") and hasattr(param_value, "n_splits"):
# CV generator (has split method and n_splits)
cv_hash = _hash_cv_generator(param_value)
key_parts.append(f"{param_name}_cv_{cv_hash}")
elif isinstance(param_value, pd.DataFrame):
df_hash = _hash_dataframe_fast(param_value)
key_parts.append(f"{param_name}_df_{df_hash}")
elif isinstance(param_value, pd.Series):
series_hash = _hash_series_fast(param_value)
key_parts.append(f"{param_name}_ser_{series_hash}")
elif isinstance(param_value, np.ndarray):
arr_hash = hashlib.md5(param_value.tobytes()).hexdigest()[:8]
key_parts.append(f"{param_name}_arr_{param_value.shape}_{arr_hash}")
elif isinstance(param_value, (str, int, float, bool)):
key_parts.append(f"{param_name}_{param_value}")
elif callable(param_value):
# For scoring functions
func_name = getattr(param_value, "__name__", str(type(param_value)))
key_parts.append(f"{param_name}_func_{func_name}")
else:
# Fallback: try to hash string representation
key_parts.append(f"{param_name}_{hash(str(param_value))}")
except Exception as e:
logger.debug(f"Failed to hash param '{param_name}': {e}")
key_parts.append(f"{param_name}_unknown")
# Create final hash
combined = "_".join(key_parts)
return hashlib.md5(combined.encode()).hexdigest()
def cv_cacheable(
_func=None,
track_data_access: bool = False,
dataset_name: Optional[str] = None,
purpose: Optional[str] = None,
log_metrics: bool = True,
):
"""
Specialized caching decorator for cross-validation functions.
Handles sklearn classifiers, CV generators, and complex ML workflows.
Dual-mode decorator that supports both old and new syntax.
# Old syntax (backward compatible)
@cv_cacheable
def my_func(...)
# New syntax
@cv_cacheable(track_data_access=True, dataset_name='my_data', purpose='train')
def my_func(...)
Args:
track_data_access: Track DataFrame access for contamination detection
dataset_name: Name of dataset for tracking
purpose: Purpose of data access (train/test/validate/optimize/analyze)
log_metrics: Log results to MLflow if available
"""
# Validate purpose parameter
if purpose and purpose not in ["train", "test", "validate", "optimize", "analyze"]:
raise ValueError(
f"Invalid purpose: {purpose}. Must be one of: "
"train, test, validate, optimize, analyze"
)
def decorator(func):
# If no enhanced parameters are set, use old behavior
if not track_data_access and dataset_name is None and purpose is None:
return _cv_cacheable_legacy(func)
else:
return _cv_cacheable_enhanced(
func,
track_data_access=track_data_access,
dataset_name=dataset_name,
purpose=purpose,
log_metrics=log_metrics,
)
if _func is None:
return decorator
else:
return decorator(_func)
def _cv_cacheable_legacy(func):
"""Original cv_cacheable implementation for backward compatibility."""
from . import CACHE_DIRS, cache_stats
func_name = f"{func.__module__}.{func.__qualname__}"
cache_dir = CACHE_DIRS["base"] / "cv_cache"
cache_dir.mkdir(exist_ok=True)
@wraps(func)
def wrapper(*args, **kwargs):
# Original cache key generation (unchanged)
cache_key = _generate_cv_cache_key(func, args, kwargs)
cache_file = cache_dir / f"{cache_key}.pkl"
if cache_file.exists():
try:
with open(cache_file, "rb") as f:
result = pickle.load(f)
cache_stats.record_hit(func_name)
logger.info(f"CV cache hit for {func.__name__}")
return result
except Exception as e:
logger.warning(f"CV cache read failed: {e}")
cache_file.unlink()
# Cache miss
cache_stats.record_miss(func_name)
result = func(*args, **kwargs)
try:
with open(cache_file, "wb") as f:
pickle.dump(result, f)
except Exception as e:
logger.warning(f"Failed to cache CV result: {e}")
return result
wrapper._afml_cacheable = True
return wrapper
def _cv_cacheable_enhanced(
func, track_data_access=False, dataset_name=None, purpose=None, log_metrics=True
):
"""Enhanced version with tracking capabilities."""
from . import CACHE_DIRS, cache_stats
from .mlflow_integration import MLFLOW_AVAILABLE, get_mlflow_cache
func_name = f"{func.__module__}.{func.__qualname__}"
cache_dir = CACHE_DIRS["base"] / "cv_cache_enhanced"
cache_dir.mkdir(exist_ok=True)
def _generate_enhanced_cv_cache_key(
base_key, track_data_access, dataset_name, purpose, log_metrics
):
"""Generate cache key that includes tracking parameters."""
tracking_params = {
"track_data_access": track_data_access,
"dataset_name": dataset_name,
"purpose": purpose,
"log_metrics": log_metrics,
}
params_hash = hashlib.md5(
json.dumps(tracking_params, sort_keys=True).encode()
).hexdigest()[:8]
return f"{base_key}_tracking_{params_hash}"
@wraps(func)
def wrapper(*args, **kwargs):
base_key = _generate_cv_cache_key(func, args, kwargs)
cache_key = _generate_enhanced_cv_cache_key(
base_key, track_data_access, dataset_name, purpose, log_metrics
)
cache_file = cache_dir / f"{cache_key}.pkl"
# Track data access - IMPORT HERE
if track_data_access:
from .data_access_tracker import get_data_tracker
_track_cv_data_access(
get_data_tracker(), args, kwargs, dataset_name, purpose
)
# Check cache
if cache_file.exists():
try:
with open(cache_file, "rb") as f:
result = pickle.load(f)
cache_stats.record_hit(func_name)
# Log cached results to MLflow
if log_metrics and MLFLOW_AVAILABLE:
_log_cv_metrics_to_mlflow(result, func_name, cache_key, "cached")
logger.info(f"Enhanced CV cache hit for {func.__name__}")
return result
except Exception as e:
logger.warning(f"Enhanced CV cache read failed: {e}")
cache_file.unlink()
# Cache miss
cache_stats.record_miss(func_name)
logger.info(f"Enhanced CV cache miss for {func.__name__} - computing...")
start_time = time.time()
result = func(*args, **kwargs)
execution_time = time.time() - start_time
# Save to enhanced cache
try:
with open(cache_file, "wb") as f:
pickle.dump(result, f)
logger.debug(f"Cached enhanced CV result: {cache_key}")
except Exception as e:
logger.warning(f"Failed to cache enhanced CV result: {e}")
# Log to MLflow
if log_metrics and MLFLOW_AVAILABLE:
_log_cv_metrics_to_mlflow(result, func_name, cache_key, "computed")
try:
mlflow_cache = get_mlflow_cache()
mlflow_cache._log_metrics({"execution_time_seconds": execution_time})
except Exception as e:
logger.debug(f"Failed to log execution time: {e}")
return result
wrapper._afml_cacheable = True
return wrapper
def _track_cv_data_access(tracker, args, kwargs, dataset_name, purpose):
"""Track data access in CV functions."""
from .robust_cache_keys import _is_trackable_dataframe
# Extract X, y from common CV function signatures
X, y = None, None
# Try to find X and y in args/kwargs
for arg in args:
if isinstance(arg, pd.DataFrame) and _is_trackable_dataframe(arg):
X = arg
elif isinstance(arg, (pd.Series, np.ndarray)) and len(arg) > 0:
y = arg
for key, value in kwargs.items():
if key in ["X", "x", "features"] and _is_trackable_dataframe(value):
X = value
elif key in ["y", "target", "labels"] and isinstance(
value, (pd.Series, np.ndarray)
):
y = value
# Log access if we found trackable data
if X is not None:
tracker.log_access(
dataset_name=dataset_name or "cv_dataset",
start_date=X.index[0],
end_date=X.index[-1],
purpose=purpose or "cv",
data_shape=X.shape,
)
def _log_cv_metrics_to_mlflow(result, func_name, cache_key):
"""Log CV metrics to MLflow for experiment tracking."""
from .mlflow_integration import get_mlflow_cache
try:
mlflow_cache = get_mlflow_cache()
with mlflow_cache.experiment_run(
run_name=f"cv_{func_name}_{cache_key[:8]}",
tags={"type": "cross_validation", "function": func_name},
) as ctx:
# Extract metrics from common CV result formats
if isinstance(result, dict):
# Direct metric dictionary
for key, value in result.items():
if isinstance(value, (int, float)):
ctx.log_metric(f"cv_{key}", value)
elif isinstance(result, (list, np.ndarray)):
# Array of scores
if len(result) > 0:
ctx.log_metric("cv_mean_score", np.mean(result))
ctx.log_metric("cv_std_score", np.std(result))
ctx.log_metric("cv_n_folds", len(result))
elif hasattr(result, "cv_results_"):
# Sklearn CV result object
for key, value in result.cv_results_.items():
if isinstance(value, (list, np.ndarray)) and len(value) > 0:
ctx.log_metric(f"cv_{key}_mean", np.mean(value))
except Exception as e:
logger.debug(f"Failed to log CV metrics to MLflow: {e}")
def clear_cv_cache():
"""Clear all CV cache files."""
from . import CACHE_DIRS
cache_dir = CACHE_DIRS["base"] / "cv_cache"
model_cache_dir = CACHE_DIRS["base"] / "cv_cache_models"
count = 0
for cache_dir in [cache_dir, model_cache_dir]:
if cache_dir.exists():
for cache_file in cache_dir.glob("*.pkl"):
cache_file.unlink()
count += 1
logger.info(f"Cleared {count} CV cache files")
return count
def cv_cache_with_classifier_state(func: Callable) -> Callable:
"""
Caching decorator that also caches the trained classifier state.
Use this if you want to cache both CV scores AND the trained models.
Returns: (original_result, cached_classifiers)
where cached_classifiers is a list of trained classifiers from each fold.
Usage:
@cv_cache_with_classifier_state
def ml_cross_val_score_with_models(classifier, X, y, cv_gen, ...):
# Your CV loop that returns (scores, trained_models)
...
"""
from . import CACHE_DIRS, cache_stats
func_name = f"{func.__module__}.{func.__qualname__}"
cache_dir = CACHE_DIRS["base"] / "cv_cache_models"
cache_dir.mkdir(exist_ok=True)
@wraps(func)
def wrapper(*args, **kwargs):
# Generate cache key
try:
cache_key = _generate_cv_cache_key(func, args, kwargs)
except Exception as e:
logger.warning(f"CV cache key generation failed: {e}")
cache_stats.record_miss(func_name)
return func(*args, **kwargs)
# Check cache
cache_file = cache_dir / f"{cache_key}.pkl"
if cache_file.exists():
try:
with open(cache_file, "rb") as f:
result = pickle.load(f)
cache_stats.record_hit(func_name)
logger.info(f"CV cache hit (with models) for {func.__name__}")
return result
except Exception as e:
logger.warning(f"CV cache read failed: {e}")
cache_file.unlink()
# Cache miss - compute
cache_stats.record_miss(func_name)
logger.info(f"CV cache miss (with models) for {func.__name__} - computing...")
result = func(*args, **kwargs)
# Save to cache
try:
with open(cache_file, "wb") as f:
pickle.dump(result, f)
logger.debug(f"Cached CV result with models: {cache_key}")
except Exception as e:
logger.warning(f"Failed to cache CV result: {e}")
return result
wrapper._afml_cacheable = True
return wrapper
__all__ = [
"cv_cacheable",
"cv_cache_with_classifier_state",
"clear_cv_cache",
]
|