#!/usr/bin/env python3 """ Generate verification results using the current model (V3) Dynamically creates verification data from the most recent matches """ import pandas as pd import numpy as np import torch import torch.nn as nn import json import os BASE_DIR = os.path.dirname(os.path.abspath(__file__)) MODELS_DIR = os.path.join(BASE_DIR, '..', 'models') DATA_DIR = os.path.join(BASE_DIR, '..', 'data') # Model definition (same as in training) class BettingNetV3(nn.Module): def __init__(self, input_size, hidden_sizes=[512, 256, 256, 128, 64], dropout=0.3): super().__init__() layers = [] prev_size = input_size for i, hidden_size in enumerate(hidden_sizes): layers.append(nn.Linear(prev_size, hidden_size)) layers.append(nn.BatchNorm1d(hidden_size)) layers.append(nn.LeakyReLU(0.1)) layers.append(nn.Dropout(dropout if i < len(hidden_sizes) - 1 else dropout * 0.5)) prev_size = hidden_size self.feature_extractor = nn.Sequential(*layers) self.classifier = nn.Linear(prev_size, 3) def forward(self, x): features = self.feature_extractor(x) return self.classifier(features) RESULT_MAP = {0: 'AWAY', 1: 'DRAW', 2: 'HOME'} LEAGUE_NAMES = { 'E0': 'Premier League', 'D1': 'Bundesliga', 'I1': 'Serie A', 'SP1': 'La Liga', 'F1': 'Ligue 1' } CORE_LEAGUES = ['E0', 'D1', 'I1', 'SP1', 'F1'] def generate_verification(matches_per_league=20): """Generate verification results from recent matches""" print("🔄 Generating verification results with V3 model...") # Load meta meta_path = os.path.join(MODELS_DIR, 'betting_model_v3_meta.json') if not os.path.exists(meta_path): print("❌ Model V3 meta not found!") return None with open(meta_path) as f: meta = json.load(f) features = meta['feature_columns'] print(f" Features: {len(features)}") # Load model model = BettingNetV3(len(features)) model.load_state_dict(torch.load( os.path.join(MODELS_DIR, 'betting_model_v3.pth'), map_location='cpu', weights_only=True )) model.eval() # Load scaler scaler_data = np.load( os.path.join(MODELS_DIR, 'scaler_v3.npy'), allow_pickle=True ).item() # Load advanced data df = pd.read_csv( os.path.join(DATA_DIR, 'advanced_historical_matches.csv'), low_memory=False ) df = df.sort_values('date').reset_index(drop=True) # Get recent matches from core leagues df_core = df[df['league_code'].isin(CORE_LEAGUES)].copy() # Take last N matches per league verification_matches = [] for league in CORE_LEAGUES: league_df = df_core[df_core['league_code'] == league].tail(matches_per_league) verification_matches.append(league_df) test_df = pd.concat(verification_matches).copy() print(f" Verification matches: {len(test_df)}") # Prepare features for col in features: if col not in test_df.columns: test_df[col] = 0 test_df[col] = test_df[col].fillna(0) X = test_df[features].values # Scale X_scaled = (X - scaler_data['mean']) / scaler_data['scale'] # Predict with torch.no_grad(): outputs = model(torch.FloatTensor(X_scaled)) probs = torch.softmax(outputs, dim=1).numpy() preds = outputs.argmax(dim=1).numpy() # Create verification results results = [] for i, (_, row) in enumerate(test_df.iterrows()): # Skip rows with missing essential data if pd.isna(row['home_team']) or pd.isna(row['away_team']) or pd.isna(row['date']): continue actual_code = int(row['result_code']) if pd.notna(row['result_code']) else 1 pred_code = int(preds[i]) # Handle potential NaN in probabilities home_prob = float(probs[i][2] * 100) if not np.isnan(probs[i][2]) else 33.3 draw_prob = float(probs[i][1] * 100) if not np.isnan(probs[i][1]) else 33.3 away_prob = float(probs[i][0] * 100) if not np.isnan(probs[i][0]) else 33.3 confidence = float(probs[i].max() * 100) if not np.isnan(probs[i].max()) else 33.3 results.append({ 'league': str(row['league_code']), 'league_name': LEAGUE_NAMES.get(row['league_code'], str(row['league_code'])), 'date': str(row['date'])[:10], 'home_team': str(row['home_team']), 'away_team': str(row['away_team']), 'home_goals': int(row['home_goals']) if pd.notna(row['home_goals']) else 0, 'away_goals': int(row['away_goals']) if pd.notna(row['away_goals']) else 0, 'actual': RESULT_MAP[actual_code], 'actual_code': actual_code, 'predicted': RESULT_MAP[pred_code], 'predicted_code': pred_code, 'confidence': confidence, 'home_prob': home_prob, 'draw_prob': draw_prob, 'away_prob': away_prob, 'correct': actual_code == pred_code }) # Save output_path = os.path.join(DATA_DIR, 'verification_results.json') with open(output_path, 'w') as f: json.dump(results, f, indent=2) # Stats correct = sum(1 for r in results if r['correct']) total = len(results) print(f"\n{'='*50}") print(f"✅ VERIFICA V3 COMPLETATA") print(f"{'='*50}") print(f" Totale partite: {total}") print(f" Corrette: {correct}") print(f" Accuracy: {100*correct/total:.1f}%") print() # Per league for league in CORE_LEAGUES: league_results = [r for r in results if r['league'] == league] if league_results: league_correct = sum(1 for r in league_results if r['correct']) print(f" {LEAGUE_NAMES[league]}: {league_correct}/{len(league_results)} ({100*league_correct/len(league_results):.0f}%)") # High confidence high_conf = [r for r in results if r['confidence'] > 50] if high_conf: high_conf_correct = sum(1 for r in high_conf if r['correct']) print(f"\n High confidence (>50%): {high_conf_correct}/{len(high_conf)} ({100*high_conf_correct/len(high_conf):.0f}%)") print(f"\n Salvato in: {output_path}") return results if __name__ == '__main__': generate_verification(matches_per_league=20)