File size: 40,990 Bytes
c49b21b |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 |
import pandas as pd
import numpy as np
from pathlib import Path
import json
import warnings
warnings.filterwarnings('ignore')
class FinalNullValueHandler:
"""
Advanced final null value handler with symbol-first temporal interpolation.
Strategy Priority:
1. Same symbol, nearby timestamps (interpolation/extrapolation)
2. Same symbol, historical mean/median
3. Similar symbols (same asset class)
4. Global defaults with symbol-specific variation
"""
def __init__(self):
self.crypto_column_defaults = self._define_crypto_defaults()
self.stock_column_defaults = self._define_stock_defaults()
self.symbol_profiles = {}
self.symbol_stats = {} # Historical statistics per symbol
def _analyze_symbol_statistics(self, df):
"""Analyze historical statistics for each symbol to guide intelligent filling"""
stats = {}
# Sort by timestamp for proper temporal analysis
if 'interval_timestamp' in df.columns:
df_sorted = df.sort_values(['symbol', 'interval_timestamp'])
else:
df_sorted = df.sort_values('symbol')
for symbol in df['symbol'].unique():
symbol_data = df_sorted[df_sorted['symbol'] == symbol].copy()
symbol_stats = {
'symbol': symbol,
'total_records': len(symbol_data),
'date_range': None,
'typical_values': {},
'volatility': {},
'trends': {},
'seasonal_patterns': {}
}
# Calculate date range if timestamp available
if 'interval_timestamp' in symbol_data.columns:
timestamps = pd.to_datetime(symbol_data['interval_timestamp'], unit='ms')
symbol_stats['date_range'] = {
'start': timestamps.min(),
'end': timestamps.max(),
'duration_days': (timestamps.max() - timestamps.min()).days
}
# Calculate typical values, volatility, and trends for numerical columns
numerical_cols = symbol_data.select_dtypes(include=[np.number]).columns
for col in numerical_cols:
if col in ['interval_timestamp', 'backup_id']:
continue
col_data = symbol_data[col].dropna()
if len(col_data) > 0:
symbol_stats['typical_values'][col] = {
'mean': col_data.mean(),
'median': col_data.median(),
'std': col_data.std(),
'min': col_data.min(),
'max': col_data.max(),
'q25': col_data.quantile(0.25),
'q75': col_data.quantile(0.75),
'recent_mean': col_data.tail(min(10, len(col_data))).mean(), # Last 10 values
'data_points': len(col_data)
}
# Calculate volatility
if len(col_data) > 1:
symbol_stats['volatility'][col] = col_data.std() / (col_data.mean() + 1e-8)
# Calculate trend if we have timestamp data
if 'interval_timestamp' in symbol_data.columns and len(col_data) >= 3:
# Simple linear trend
valid_rows = symbol_data[col].notna()
if valid_rows.sum() >= 3:
x = np.arange(len(symbol_data[valid_rows]))
y = symbol_data.loc[valid_rows, col].values
try:
trend_slope = np.polyfit(x, y, 1)[0]
symbol_stats['trends'][col] = trend_slope
except:
symbol_stats['trends'][col] = 0
stats[symbol] = symbol_stats
return stats
def _temporal_interpolation_fill(self, df, symbol, column):
"""
Fill nulls using temporal interpolation within the same symbol
Priority:
1. Linear interpolation between known values
2. Forward fill from last known value
3. Backward fill from next known value
4. Exponential smoothing for trend continuation
"""
try:
symbol_mask = df['symbol'] == symbol
symbol_data = df.loc[symbol_mask].copy()
if column not in symbol_data.columns or symbol_data[column].notna().sum() == 0:
return None
# Sort by timestamp if available and remove duplicates
if 'interval_timestamp' in symbol_data.columns:
symbol_data = symbol_data.sort_values('interval_timestamp')
# Drop duplicate timestamps for this symbol to avoid reindex issues
symbol_data = symbol_data.drop_duplicates(subset=['interval_timestamp'], keep='first')
# Reset index to avoid any index issues
symbol_data = symbol_data.reset_index(drop=True)
filled_series = symbol_data[column].copy()
# 1. Linear interpolation (works best with timestamp ordering)
if 'interval_timestamp' in symbol_data.columns and len(symbol_data) > 1:
# Try time-based interpolation with safe fallback
try:
original_index = filled_series.index
datetime_index = pd.to_datetime(symbol_data['interval_timestamp'], unit='ms')
# Ensure unique datetime index
if datetime_index.duplicated().any():
# Add microseconds to make unique
for i, is_dup in enumerate(datetime_index.duplicated(keep='first')):
if is_dup:
datetime_index.iloc[i] += pd.Timedelta(microseconds=i+1)
filled_series.index = datetime_index
filled_series = filled_series.interpolate(method='time')
filled_series.index = original_index # Restore original index
except Exception:
# Fallback to linear interpolation if time interpolation fails
filled_series = filled_series.interpolate(method='linear')
else:
filled_series = filled_series.interpolate(method='linear')
# 2. Forward fill
filled_series = filled_series.ffill()
# 3. Backward fill
filled_series = filled_series.bfill()
# 4. If still has nulls, use trend extrapolation
if filled_series.isna().any() and symbol in self.symbol_stats:
symbol_stat = self.symbol_stats[symbol]
if column in symbol_stat.get('typical_values', {}):
typical_val = symbol_stat['typical_values'][column]['recent_mean']
trend = symbol_stat.get('trends', {}).get(column, 0)
# Apply trend-based extrapolation for remaining nulls
for idx in filled_series[filled_series.isna()].index:
# Simple trend continuation
filled_series[idx] = typical_val + trend * (idx % 10) # Modest trend application
return filled_series
except Exception as e:
# If all else fails, return None to trigger fallback behavior
print(f"Warning: Temporal interpolation failed for {symbol} {column}: {e}")
return None
def _similar_symbol_fill(self, df, symbol, column, asset_type):
"""
Fill nulls using similar symbols in the same asset class
"""
if asset_type == 'crypto':
# For crypto, use symbols with similar rank or market cap
target_stats = self.symbol_stats.get(symbol, {})
target_rank = target_stats.get('typical_values', {}).get('rank', {}).get('median', 999)
similar_symbols = []
for sym, stats in self.symbol_stats.items():
if sym == symbol:
continue
sym_rank = stats.get('typical_values', {}).get('rank', {}).get('median', 999)
if abs(sym_rank - target_rank) <= 50: # Similar rank range
similar_symbols.append(sym)
else: # stock
# For stocks, use symbols with similar market cap or sector
target_stats = self.symbol_stats.get(symbol, {})
target_mcap = target_stats.get('typical_values', {}).get('marketCapitalization', {}).get('median', 0)
similar_symbols = []
for sym, stats in self.symbol_stats.items():
if sym == symbol:
continue
sym_mcap = stats.get('typical_values', {}).get('marketCapitalization', {}).get('median', 0)
if target_mcap > 0 and sym_mcap > 0:
ratio = max(sym_mcap, target_mcap) / min(sym_mcap, target_mcap)
if ratio <= 5: # Within 5x market cap
similar_symbols.append(sym)
if not similar_symbols:
return None
# Get values from similar symbols
similar_data = df[df['symbol'].isin(similar_symbols)][column].dropna()
if len(similar_data) > 0:
# Use weighted average based on similarity
return similar_data.median() # Robust central tendency
return None
def _intelligent_symbol_fill(self, df, symbol, column):
"""
Intelligent filling strategy prioritizing symbol-specific data
Returns the best estimate for null values in the specified column for the given symbol
"""
# Strategy 1: Temporal interpolation within same symbol
temporal_result = self._temporal_interpolation_fill(df, symbol, column)
if temporal_result is not None and temporal_result.notna().any():
return temporal_result
# Strategy 2: Use historical statistics from same symbol
if symbol in self.symbol_stats and column in self.symbol_stats[symbol]['typical_values']:
stats = self.symbol_stats[symbol]['typical_values'][column]
# Choose appropriate central tendency based on data characteristics
if stats['data_points'] >= 10:
# Use recent mean for frequently updated data
return stats['recent_mean']
elif stats['data_points'] >= 3:
# Use median for small datasets (more robust)
return stats['median']
else:
# Use mean for very small datasets
return stats['mean']
# Strategy 3: Use similar symbols
asset_type = 'crypto' if symbol in df.columns and any(
col in df.columns for col in ['rank', 'dominance', 'performance.day']
) else 'stock'
similar_fill = self._similar_symbol_fill(df, symbol, column, asset_type)
if similar_fill is not None:
return similar_fill
# Strategy 4: Global fallback with symbol variation
return None # Will be handled by existing default logic
def _define_crypto_defaults(self):
"""Define intelligent defaults for crypto-specific columns"""
return {
# Crypto market data
'dominance': 0.001, # Very small dominance for minor cryptos
'rank': 999, # Low rank for unknown cryptos
'stable': 0, # Most cryptos are not stablecoins (use 0 instead of False)
'marketcap': 1000000, # $1M default market cap
'transaction_count': 100, # Minimal transaction count
'transaction_volume': 10000, # Minimal transaction volume
'tx_price_correlation': 0.5, # Neutral correlation
# Exchange prices (use main price as baseline)
'exchangePrices.binance': None, # Will be filled with main price
'exchangePrices.coinbase': None,
'exchangePrices.kraken': None,
'exchangePrices.bybit': None,
'exchangePrices.kucoin': None,
'exchangePrices.okx': None,
'exchangePrices.mexc': None,
'exchangePrices.gateio': None,
'exchangePrices.bitget': None,
'exchangePrices.bitmart': None,
'exchangePrices.bingx': None,
'exchangePrices.cryptocom': None,
# Exchange symbols (use main symbol as baseline)
'symbols.binance': None, # Will be filled with main symbol
'symbols.coinbase': None,
'symbols.kraken': None,
'symbols.bybit': None,
'symbols.kucoin': None,
'symbols.okx': None,
'symbols.mexc': None,
'symbols.gateio': None,
'symbols.bitget': None,
'symbols.bitmart': None,
'symbols.bingx': None,
'symbols.cryptocom': None,
# Performance metrics (neutral/small changes)
'performance.day': 0.0,
'performance.hour': 0.0,
'performance.hour4': 0.0,
'performance.min1': 0.0,
'performance.min15': 0.0,
'performance.min5': 0.0,
'performance.month': 0.0,
'performance.month3': 0.0,
'performance.week': 0.0,
'performance.year': 0.0,
# Rank differences (no change)
'rankDiffs.day': 0,
'rankDiffs.hour': 0,
'rankDiffs.hour4': 0,
'rankDiffs.min1': 0,
'rankDiffs.min15': 0,
'rankDiffs.min5': 0,
'rankDiffs.month': 0,
'rankDiffs.month3': 0,
'rankDiffs.week': 0,
'rankDiffs.year': 0,
# Technical indicators
'bb_width': 0.02, # Small bollinger band width
'cg_id': None, # Will be derived from symbol
}
def _define_stock_defaults(self):
"""Define intelligent defaults for stock-specific columns"""
return {
# Stock market data
'stock_market': 'NASDAQ', # Default market
'marketCapitalization': 1000000000, # $1B default
'shareOutstanding': 100000000, # 100M shares default
'mspr': 0, # Neutral momentum
# News and sentiment data
'news_activity_score_x': 0,
'news_activity_score_y': 0,
'news_articles_count_x': 0,
'news_articles_count_y': 0,
'news_highlights_count_x': 0,
'news_highlights_count_y': 0,
'news_match_score_max_x': 0,
'news_match_score_max_y': 0,
'news_match_score_mean_x': 0,
'news_match_score_mean_y': 0,
'news_mentions_count_x': 0,
'news_mentions_count_y': 0,
'news_sentiment_max_x': 0.5, # Neutral sentiment
'news_sentiment_max_y': 0.5,
'news_sentiment_mean_x': 0.5,
'news_sentiment_mean_y': 0.5,
'news_sentiment_min_x': 0.5,
'news_sentiment_min_y': 0.5,
'news_sentiment_range_x': 0,
'news_sentiment_range_y': 0,
'news_sentiment_std': 0,
'news_sentiment_std_x': 0,
'news_sentiment_std_y': 0,
# Analyst ratings
'buy': 5, # Moderate buy recommendations
'hold': 10, # More hold recommendations
'sell': 2, # Few sell recommendations
'strongBuy': 3,
'strongSell': 1,
# Technical indicators
'volume_price_momentum': 0.0, # Neutral momentum
}
def _create_symbol_profiles(self, df):
"""Create profiles for each symbol to guide intelligent filling"""
profiles = {}
for symbol in df['symbol'].unique():
symbol_data = df[df['symbol'] == symbol]
# Determine if it's crypto or stock
is_crypto = 'rank' in symbol_data.columns and symbol_data['rank'].notna().any()
if not is_crypto:
is_crypto = any(col.startswith('performance.') for col in symbol_data.columns)
# Calculate key statistics
profile = {
'symbol': symbol,
'is_crypto': is_crypto,
'total_records': len(symbol_data),
'data_density': symbol_data.notna().mean().mean(),
'has_price_data': 'price' in symbol_data.columns and symbol_data['price'].notna().any(),
'typical_price': symbol_data.get('price', pd.Series([100])).median(),
'typical_volume': symbol_data.get('volume', pd.Series([1000000])).median(),
'typical_marketcap': symbol_data.get('marketcap', symbol_data.get('marketCapitalization', pd.Series([1000000000]))).median()
}
profiles[symbol] = profile
return profiles
def _intelligent_fill_value(self, df, symbol, column, default_value):
"""Generate intelligent fill value based on symbol context"""
profile = self.symbol_profiles.get(symbol, {})
# Add symbol-specific variation to prevent homogenization
symbol_hash = hash(f"{symbol}_{column}") % 1000
variation_factor = (symbol_hash / 1000.0 - 0.5) * 0.1 # Β±5% variation
if default_value is None:
return None
elif isinstance(default_value, (int, float)):
if default_value == 0:
return 0 # Keep zeros as zeros
else:
return default_value * (1 + variation_factor)
else:
return default_value
def _fill_exchange_prices_advanced(self, df):
"""Advanced exchange price filling using symbol-first strategy"""
exchange_price_cols = [col for col in df.columns if col.startswith('exchangePrices.')]
if not exchange_price_cols or 'price' not in df.columns:
return df
df_result = df.copy()
for symbol in df['symbol'].unique():
symbol_mask = df['symbol'] == symbol
symbol_data = df.loc[symbol_mask]
# First try to get main price from symbol's own data
main_price_series = self._intelligent_symbol_fill(df, symbol, 'price')
if main_price_series is None or (isinstance(main_price_series, pd.Series) and main_price_series.isna().all()):
continue
if isinstance(main_price_series, pd.Series):
main_price = main_price_series.median()
else:
main_price = main_price_series
if pd.isna(main_price):
continue
# Fill exchange prices for this symbol
for exchange_col in exchange_price_cols:
if symbol_data[exchange_col].isna().any():
# First try temporal interpolation for this exchange
exchange_filled = self._intelligent_symbol_fill(df, symbol, exchange_col)
if exchange_filled is not None:
if isinstance(exchange_filled, pd.Series):
df_result.loc[symbol_mask, exchange_col] = exchange_filled
else:
null_mask = df_result.loc[symbol_mask, exchange_col].isna()
df_result.loc[symbol_mask & null_mask, exchange_col] = exchange_filled
else:
# Fallback: use main price with small exchange-specific variation
exchange_hash = hash(f"{symbol}_{exchange_col}") % 100
variation = (exchange_hash / 100.0 - 0.5) * 0.01 # Β±0.5%
exchange_price = main_price * (1 + variation)
null_mask = df_result.loc[symbol_mask, exchange_col].isna()
df_result.loc[symbol_mask & null_mask, exchange_col] = exchange_price
return df_result
def _fill_exchange_symbols(self, df):
"""Fill exchange symbols with main symbol + exchange-specific formatting"""
exchange_symbol_cols = [col for col in df.columns if col.startswith('symbols.')]
if not exchange_symbol_cols or 'symbol' not in df.columns:
return df
df_result = df.copy()
# Exchange-specific symbol formatting
exchange_formats = {
'symbols.binance': lambda s: f"{s.upper()}USDT" if s.lower() != 'bitcoin' else "BTCUSDT",
'symbols.coinbase': lambda s: f"{s.upper()}-USD",
'symbols.kraken': lambda s: f"{s.upper()}USD" if len(s) <= 3 else f"{s.upper()}/USD",
'symbols.bybit': lambda s: f"{s.upper()}USDT",
'symbols.kucoin': lambda s: f"{s.upper()}-USDT",
'symbols.okx': lambda s: f"{s.upper()}-USDT",
'symbols.mexc': lambda s: f"{s.upper()}_USDT",
'symbols.gateio': lambda s: f"{s.upper()}_USDT",
'symbols.bitget': lambda s: f"{s.upper()}USDT",
'symbols.bitmart': lambda s: f"{s.upper()}_USDT",
'symbols.bingx': lambda s: f"{s.upper()}-USDT",
'symbols.cryptocom': lambda s: f"{s.upper()}_USDT"
}
for symbol in df['symbol'].unique():
symbol_mask = df['symbol'] == symbol
for exchange_col in exchange_symbol_cols:
if df.loc[symbol_mask, exchange_col].isna().all():
formatter = exchange_formats.get(exchange_col, lambda s: s.upper())
try:
exchange_symbol = formatter(symbol)
df_result.loc[symbol_mask, exchange_col] = exchange_symbol
except Exception:
df_result.loc[symbol_mask, exchange_col] = symbol.upper()
return df_result
def _fill_cg_id(self, df):
"""Fill CoinGecko ID based on symbol"""
if 'cg_id' not in df.columns:
return df
df_result = df.copy()
# Common CoinGecko ID mappings
cg_id_mapping = {
'bitcoin': 'bitcoin',
'btc': 'bitcoin',
'ethereum': 'ethereum',
'eth': 'ethereum',
'binancecoin': 'binancecoin',
'bnb': 'binancecoin',
'cardano': 'cardano',
'ada': 'cardano',
'solana': 'solana',
'sol': 'solana',
'xrp': 'ripple',
'ripple': 'ripple',
'dogecoin': 'dogecoin',
'doge': 'dogecoin',
'polkadot': 'polkadot',
'dot': 'polkadot',
'avalanche-2': 'avalanche-2',
'avax': 'avalanche-2',
'chainlink': 'chainlink',
'link': 'chainlink',
'polygon': 'matic-network',
'matic': 'matic-network',
'litecoin': 'litecoin',
'ltc': 'litecoin',
'uniswap': 'uniswap',
'uni': 'uniswap'
}
for symbol in df['symbol'].unique():
symbol_mask = df['symbol'] == symbol
if df.loc[symbol_mask, 'cg_id'].isna().all():
cg_id = cg_id_mapping.get(symbol.lower(), symbol.lower())
df_result.loc[symbol_mask, 'cg_id'] = cg_id
return df_result
def process_crypto_features(self, df):
"""Process crypto features with advanced symbol-first null handling"""
print("Processing crypto features with symbol-first strategy...")
df_result = df.copy()
# Step 1: Analyze symbol statistics for intelligent filling
print("Analyzing symbol statistics...")
self.symbol_stats = self._analyze_symbol_statistics(df_result)
print(f"Analyzed {len(self.symbol_stats)} symbols")
# Step 2: Create symbol profiles
self.symbol_profiles = self._create_symbol_profiles(df_result)
# Step 3: Symbol-first null handling for key columns
priority_columns = [
'price', 'volume', 'marketcap', 'dominance', 'rank',
'performance.day', 'performance.week', 'performance.month',
'rsi', 'macd', 'transaction_count', 'transaction_volume'
]
for column in priority_columns:
if column in df_result.columns and df_result[column].isna().any():
print(f"Processing {column} with symbol-first strategy...")
for symbol in df_result['symbol'].unique():
symbol_mask = df_result['symbol'] == symbol
null_mask = df_result[column].isna()
fill_mask = symbol_mask & null_mask
if fill_mask.any():
# Use intelligent symbol-first filling
fill_result = self._intelligent_symbol_fill(df_result, symbol, column)
if fill_result is not None:
if isinstance(fill_result, pd.Series):
# If we got a series back (from temporal interpolation)
# Make sure the series aligns with the symbol mask
symbol_indices = df_result[symbol_mask].index
if len(fill_result) == len(symbol_indices):
# Map the series values to the correct indices
for i, idx in enumerate(symbol_indices):
if pd.notna(fill_result.iloc[i]):
df_result.loc[idx, column] = fill_result.iloc[i]
else:
# Fallback: use median of the series
fill_value = fill_result.median()
if pd.notna(fill_value):
df_result.loc[fill_mask, column] = fill_value
else:
# If we got a scalar value
df_result.loc[fill_mask, column] = fill_result
# Step 4: Handle exchange prices with cross-reference to main price
df_result = self._fill_exchange_prices_advanced(df_result)
# Step 5: Handle exchange symbols with proper formatting
df_result = self._fill_exchange_symbols(df_result)
# Step 6: Handle CoinGecko IDs
df_result = self._fill_cg_id(df_result)
# Step 7: Fill remaining columns with intelligent defaults
for column in df_result.columns:
if df_result[column].isna().any():
default_value = self.crypto_column_defaults.get(column)
if default_value is not None:
for symbol in df_result['symbol'].unique():
symbol_mask = df_result['symbol'] == symbol
null_mask = df_result[column].isna()
fill_mask = symbol_mask & null_mask
if fill_mask.any():
try:
fill_value = self._intelligent_fill_value(
df_result, symbol, column, default_value
)
df_result.loc[fill_mask, column] = fill_value
except Exception as e:
print(f"Warning: Failed to fill {column} for {symbol}: {e}")
# Skip this column for this symbol
continue
return df_result
def process_stock_features(self, df):
"""Process stock features with advanced symbol-first null handling"""
print("Processing stock features with symbol-first strategy...")
df_result = df.copy()
# Step 1: Analyze symbol statistics for intelligent filling
print("Analyzing symbol statistics...")
self.symbol_stats = self._analyze_symbol_statistics(df_result)
print(f"Analyzed {len(self.symbol_stats)} symbols")
# Step 2: Create symbol profiles
self.symbol_profiles = self._create_symbol_profiles(df_result)
# Step 3: Symbol-first null handling for key columns
priority_columns = [
'close', 'open', 'high', 'low', 'volume', 'prev_close',
'marketCapitalization', 'shareOutstanding',
'rsi', 'macd', 'atr', 'bb_position',
'news_sentiment_mean_x', 'news_sentiment_mean_y',
'buy', 'sell', 'hold', 'strongBuy', 'strongSell'
]
for column in priority_columns:
if column in df_result.columns and df_result[column].isna().any():
print(f"Processing {column} with symbol-first strategy...")
for symbol in df_result['symbol'].unique():
symbol_mask = df_result['symbol'] == symbol
null_mask = df_result[column].isna()
fill_mask = symbol_mask & null_mask
if fill_mask.any():
# Use intelligent symbol-first filling
fill_result = self._intelligent_symbol_fill(df_result, symbol, column)
if fill_result is not None:
if isinstance(fill_result, pd.Series):
# If we got a series back (from temporal interpolation)
# Make sure the series aligns with the symbol mask
symbol_indices = df_result[symbol_mask].index
if len(fill_result) == len(symbol_indices):
# Map the series values to the correct indices
for i, idx in enumerate(symbol_indices):
if pd.notna(fill_result.iloc[i]):
df_result.loc[idx, column] = fill_result.iloc[i]
else:
# Fallback: use median of the series
fill_value = fill_result.median()
if pd.notna(fill_value):
df_result.loc[fill_mask, column] = fill_value
else:
# If we got a scalar value
df_result.loc[fill_mask, column] = fill_result
# Step 4: Fill remaining columns with intelligent defaults
for column in df_result.columns:
if df_result[column].isna().any():
default_value = self.stock_column_defaults.get(column)
if default_value is not None:
for symbol in df_result['symbol'].unique():
symbol_mask = df_result['symbol'] == symbol
null_mask = df_result[column].isna()
fill_mask = symbol_mask & null_mask
if fill_mask.any():
try:
fill_value = self._intelligent_fill_value(
df_result, symbol, column, default_value
)
df_result.loc[fill_mask, column] = fill_value
except Exception as e:
print(f"Warning: Failed to fill {column} for {symbol}: {e}")
# Skip this column for this symbol
continue
return df_result
def generate_report(self, df_before, df_after, feature_type):
"""Generate a comprehensive report of null value handling with symbol-first strategy details"""
before_nulls = df_before.isnull().sum()
after_nulls = df_after.isnull().sum()
null_reduction = before_nulls - after_nulls
columns_fixed = null_reduction[null_reduction > 0]
# Analyze symbol coverage
symbol_analysis = {}
if 'symbol' in df_before.columns:
for symbol in df_before['symbol'].unique():
symbol_before = int(df_before[df_before['symbol'] == symbol].isnull().sum().sum())
symbol_after = int(df_after[df_after['symbol'] == symbol].isnull().sum().sum())
symbol_analysis[symbol] = {
'nulls_before': symbol_before,
'nulls_after': symbol_after,
'nulls_filled': symbol_before - symbol_after,
'records': int(len(df_before[df_before['symbol'] == symbol]))
}
# Analyze temporal coverage if timestamp available
temporal_analysis = {}
if 'interval_timestamp' in df_before.columns:
df_before_ts = df_before.copy()
df_after_ts = df_after.copy()
df_before_ts['date'] = pd.to_datetime(df_before_ts['interval_timestamp'], unit='ms').dt.date
df_after_ts['date'] = pd.to_datetime(df_after_ts['interval_timestamp'], unit='ms').dt.date
for date in df_before_ts['date'].unique():
date_before = int(df_before_ts[df_before_ts['date'] == date].isnull().sum().sum())
date_after = int(df_after_ts[df_after_ts['date'] == date].isnull().sum().sum())
temporal_analysis[str(date)] = {
'nulls_before': date_before,
'nulls_after': date_after,
'nulls_filled': date_before - date_after
}
report = {
'feature_type': feature_type,
'timestamp': pd.Timestamp.now().isoformat(),
'strategy': 'symbol-first-temporal-interpolation',
'total_rows': int(len(df_after)),
'total_columns': int(len(df_after.columns)),
'unique_symbols': int(len(df_after['symbol'].unique())) if 'symbol' in df_after.columns else 0,
'columns_with_nulls_before': int((before_nulls > 0).sum()),
'columns_with_nulls_after': int((after_nulls > 0).sum()),
'total_nulls_before': int(before_nulls.sum()),
'total_nulls_after': int(after_nulls.sum()),
'total_nulls_filled': int(null_reduction.sum()),
'columns_fixed': int(len(columns_fixed)),
'null_reduction_rate': float((null_reduction.sum() / before_nulls.sum()) if before_nulls.sum() > 0 else 0),
'remaining_null_columns': {str(k): int(v) for k, v in after_nulls[after_nulls > 0].to_dict().items()},
'fixed_columns_detail': {str(k): int(v) for k, v in null_reduction[null_reduction > 0].to_dict().items()},
'symbol_analysis': symbol_analysis,
'temporal_analysis': temporal_analysis,
'strategy_details': {
'symbol_stats_analyzed': len(self.symbol_stats),
'temporal_interpolation_used': True,
'similar_symbol_fallback': True,
'intelligent_defaults': True
}
}
return report
def process_crypto_features_file(input_path, output_path=None):
"""Process crypto features file"""
if output_path is None:
output_path = input_path
print(f"Loading crypto features from {input_path}...")
df = pd.read_parquet(input_path)
print(f"Loaded {len(df)} rows with {len(df.columns)} columns")
print(f"Null values before processing: {df.isnull().sum().sum()}")
handler = FinalNullValueHandler()
df_processed = handler.process_crypto_features(df)
print(f"Null values after processing: {df_processed.isnull().sum().sum()}")
# Generate report
report = handler.generate_report(df, df_processed, 'crypto')
# Save processed data
df_processed.to_parquet(output_path, index=False)
print(f"Saved processed crypto features to {output_path}")
# Save report
report_path = str(output_path).replace('.parquet', '_null_handling_report.json')
with open(report_path, 'w') as f:
json.dump(report, f, indent=2)
print(f"Saved processing report to {report_path}")
return df_processed, report
def process_stock_features_file(input_path, output_path=None):
"""Process stock features file"""
if output_path is None:
output_path = input_path
print(f"Loading stock features from {input_path}...")
df = pd.read_parquet(input_path)
print(f"Loaded {len(df)} rows with {len(df.columns)} columns")
print(f"Null values before processing: {df.isnull().sum().sum()}")
handler = FinalNullValueHandler()
df_processed = handler.process_stock_features(df)
print(f"Null values after processing: {df_processed.isnull().sum().sum()}")
# Generate report
report = handler.generate_report(df, df_processed, 'stock')
# Save processed data
df_processed.to_parquet(output_path, index=False)
print(f"Saved processed stock features to {output_path}")
# Save report
report_path = str(output_path).replace('.parquet', '_null_handling_report.json')
with open(report_path, 'w') as f:
json.dump(report, f, indent=2)
print(f"Saved processing report to {report_path}")
return df_processed, report
def main():
"""Main function to process both crypto and stock features"""
crypto_path = Path("data/merged/features/crypto_features.parquet")
stocks_path = Path("data/merged/features/stocks_features.parquet")
processed_files = []
# Process crypto features
if crypto_path.exists():
try:
df_crypto, report_crypto = process_crypto_features_file(crypto_path)
processed_files.append(('crypto', crypto_path, report_crypto))
print(f"β Crypto features processed: {report_crypto['total_nulls_filled']} nulls filled")
except Exception as e:
print(f"β Error processing crypto features: {e}")
else:
print(f"Warning: {crypto_path} not found")
# Process stock features
if stocks_path.exists():
try:
df_stocks, report_stocks = process_stock_features_file(stocks_path)
processed_files.append(('stocks', stocks_path, report_stocks))
print(f"β Stock features processed: {report_stocks['total_nulls_filled']} nulls filled")
except Exception as e:
print(f"β Error processing stock features: {e}")
else:
print(f"Warning: {stocks_path} not found")
# Summary report
if processed_files:
print("\n" + "="*60)
print("FINAL NULL VALUE HANDLING SUMMARY")
print("="*60)
total_nulls_filled = 0
for file_type, file_path, report in processed_files:
total_nulls_filled += report['total_nulls_filled']
print(f"\n{file_type.upper()} FEATURES:")
print(f" File: {file_path}")
print(f" Rows: {report['total_rows']:,}")
print(f" Columns: {report['total_columns']}")
print(f" Nulls filled: {report['total_nulls_filled']:,}")
print(f" Columns fixed: {report['columns_fixed']}")
print(f" Remaining null columns: {len(report['remaining_null_columns'])}")
if report['remaining_null_columns']:
print(f" Still have nulls: {list(report['remaining_null_columns'].keys())}")
print(f"\nTOTAL NULLS FILLED ACROSS ALL FILES: {total_nulls_filled:,}")
print("="*60)
else:
print("No files were processed successfully.")
if __name__ == "__main__":
main()
|