Spaces:
Runtime error
Runtime error
File size: 11,302 Bytes
f4481f7 | 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 | import os
import time
import json
import pickle
from datetime import datetime
import numpy as np
import pandas as pd
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier, StackingClassifier
from sklearn.model_selection import GridSearchCV, RandomizedSearchCV, cross_val_score
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, classification_report
from src.utils.logger import get_logger
from src.utils.state import TrainingState
from src.config.config import Config, ModelConfig
logger = get_logger(__name__)
class ModelTraining:
def __init__(self):
self.config = Config()
self.param_grids = ModelConfig.models
def save_pickle_files(self, state: TrainingState):
try:
timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
output_dir = os.path.join(self.config.OUTPUT_BASE_DIR, timestamp)
models_dir = os.path.join(output_dir, "models")
observations_dir = os.path.join(output_dir, "observations")
os.makedirs(models_dir, exist_ok=True)
os.makedirs(observations_dir, exist_ok=True)
vectorizer_path = os.path.join(models_dir, "vectorizer.pkl")
with open(vectorizer_path, 'wb') as f:
pickle.dump(state.tfidf_vectorizer, f)
logger.info(f"Saved TF-IDF vectorizer: {vectorizer_path}")
best_model_path = os.path.join(models_dir, f"{state.best_model_name}_model.pkl")
with open(best_model_path, 'wb') as f:
pickle.dump(state.best_model, f)
logger.info(f"Saved best model: {state.best_model_name}_model.pkl")
metadata = {
'timestamp': timestamp,
'best_model_name': state.best_model_name,
'best_model_params': str(state.best_params),
'best_model_metrics': str(state.model_metrics[state.best_model_name]),
'all_models': ', '.join(list(state.trained_models.keys())),
'tfidf_features': state.X_train_tfidf.shape[1],
'train_samples': len(state.y_train),
'test_samples': len(state.y_test)
}
metadata_path = os.path.join(observations_dir, "model_metadata.csv")
pd.DataFrame([metadata]).to_csv(metadata_path, index=False)
logger.info(f"Saved metadata: {metadata_path}")
return output_dir
except Exception as e:
logger.error(f"Failed to save pickle files: {str(e)}")
raise
def save_metrics_to_csv(self, state: TrainingState, output_dir: str):
observations_dir = os.path.join(output_dir, "observations")
os.makedirs(observations_dir, exist_ok=True)
# 1. Model Comparison Summary
# ----------------------------------------------------------------------
metrics_data = []
for model_name, metrics in state.model_metrics.items():
metrics_data.append({
'Model': model_name,
'Accuracy': metrics['accuracy'],
'Precision': metrics['precision'],
'Recall': metrics['recall'],
'F1_Score': metrics['f1_score'],
'CV_Score': metrics.get('best_cv_score', 'N/A'),
'Is_Best_Model': '1' if model_name == state.best_model_name else '0'
})
df_summary = pd.DataFrame(metrics_data)
df_summary = df_summary.sort_values('Accuracy', ascending=False)
summary_path = os.path.join(observations_dir, "model_comparison_summary.csv")
df_summary.to_csv(summary_path, index=False)
logger.info(f"Saved: model_comparison_summary.csv")
# 2. Best Parameters for Each Model
# ----------------------------------------------------------------------
params_data = []
for model_name, metrics in state.model_metrics.items():
best_params = metrics.get('best_params', {})
if isinstance(best_params, dict):
params_str = json.dumps(best_params, indent=2)
else:
params_str = str(best_params)
params_data.append({
'Model': model_name,
'Best_Parameters': params_str,
'CV_Score': metrics.get('best_cv_score', 'N/A')
})
df_params = pd.DataFrame(params_data)
params_path = os.path.join(observations_dir, "best_parameters.csv")
df_params.to_csv(params_path, index=False)
logger.info(f"Saved: best_parameters.csv")
# 3. Cross-Validation Results Summary
# ----------------------------------------------------------------------
if state.cv_results:
cv_summary = []
for model_name, cv_data in state.cv_results.items():
cv_summary.append({
'Model': model_name,
'Best_CV_Score': cv_data.get('best_score', 'N/A'),
'Best_Parameters': json.dumps(cv_data.get('best_params', {}))
})
df_cv = pd.DataFrame(cv_summary)
cv_path = os.path.join(observations_dir, "cross_validation_summary.csv")
df_cv.to_csv(cv_path, index=False)
logger.info(f"Saved: cross_validation_summary.csv")
# 4. Best Model Information
# ----------------------------------------------------------------------
best_model_info = {
'Attribute': [
'Best Model Name',
'Accuracy',
'Precision',
'Recall',
'F1-Score',
'CV Score',
'Best Parameters'
],
'Value': [
state.best_model_name,
state.model_metrics[state.best_model_name]['accuracy'],
state.model_metrics[state.best_model_name]['precision'],
state.model_metrics[state.best_model_name]['recall'],
state.model_metrics[state.best_model_name]['f1_score'],
state.model_metrics[state.best_model_name].get('best_cv_score', 'N/A'),
json.dumps(state.best_params, indent=2) if isinstance(state.best_params, dict) else str(state.best_params)
]
}
df_best = pd.DataFrame(best_model_info)
best_path = os.path.join(observations_dir, "best_model_info.csv")
df_best.to_csv(best_path, index=False)
logger.info(f"Saved: best_model_info.csv")
def train_models(self, state: TrainingState, cv_folds: int = 5) -> TrainingState:
logger.info("Model training started")
logger.info(f"Using GridSearchCV with {cv_folds}-fold CV")
try:
X_train = state.X_train_tfidf
X_test = state.X_test_tfidf
y_train = state.y_train
y_test = state.y_test
trained_models, model_metrics, cv_results = {}, {}, {}
# Define model instances
models = {
'LogisticRegression': LogisticRegression(random_state=42),
'DecisionTree': DecisionTreeClassifier(random_state=42),
'SVM': SVC(random_state=42),
'KNN': KNeighborsClassifier(),
'RandomForest': RandomForestClassifier(random_state=42)
}
for model_name, model in models.items():
start_time = time.time()
logger.info(f"\n{'='*60}")
logger.info(f"Training {model_name}...")
param_grid = self.param_grids.get(model_name, {})
search = GridSearchCV(model,
param_grid=param_grid,
cv=cv_folds,
scoring='f1',
n_jobs=-1
)
search.fit(X_train, y_train)
best_model = search.best_estimator_
y_pred = best_model.predict(X_test)
metrics = {
'accuracy': accuracy_score(y_test, y_pred),
'precision': precision_score(y_test, y_pred, average='weighted', zero_division=0),
'recall': recall_score(y_test, y_pred, average='weighted', zero_division=0),
'f1_score': f1_score(y_test, y_pred, average='weighted', zero_division=0),
'best_params': search.best_params_,
'best_cv_score': search.best_score_
}
trained_models[model_name] = best_model
model_metrics[model_name] = metrics
cv_results[model_name] = {
'cv_scores': search.cv_results_,
'best_params': search.best_params_,
'best_score': search.best_score_
}
end_time = time.time()
logger.info(f"{model_name} - Training time: {end_time - start_time:.2f} seconds")
logger.info(f"{model_name} - Best Parameters: {search.best_params_}")
logger.info(f"{model_name} - CV Score: {search.best_score_:.4f}")
logger.info(f"{model_name} - Test Accuracy: {metrics['accuracy']:.4f}")
logger.info(f"{model_name} - Test Precision: {metrics['precision']:.4f}")
logger.info(f"{model_name} - Test Recall: {metrics['recall']:.4f}")
logger.info(f"{model_name} - Test F1-Score: {metrics['f1_score']:.4f}")
# Find best model based on F1-score
best_model_name = max(model_metrics, key=lambda x: model_metrics[x]['f1_score'])
best_model = trained_models[best_model_name]
best_params = model_metrics[best_model_name]['best_params']
logger.info(f"{'='*60}")
logger.info(f"BEST MODEL: {best_model_name}")
logger.info(f"Best F1-Score: {model_metrics[best_model_name]['f1_score']:.4f}")
logger.info(f"Best Parameters: {best_params}")
logger.info(f"{'='*60}")
state.trained_models = trained_models
state.model_metrics = model_metrics
state.best_model_name = best_model_name
state.best_model = best_model
state.best_params = best_params
state.cv_results = cv_results
output_dir = self.save_pickle_files(state)
self.save_metrics_to_csv(state, output_dir)
logger.info("\nModel training completed successfully")
logger.info(f"All outputs saved to: {output_dir}/")
return state
except Exception as e:
logger.error(f"Failed to train models: {str(e)}")
raise e |