Sistema de MonitoreoFarmacéutico Industrial

1class AnomalyDetectionEnsemble:
2    """Ensemble de modelos para detección de anomalías."""
3    
4    def __init__(self, contamination: float = 0.05):
5        self.models = {
6            'isolation_forest': IsolationForest(
7                contamination=contamination,
8                n_estimators=100,
9                random_state=42
10            ),
11            'local_outlier_factor': LocalOutlierFactor(
12                n_neighbors=20,
13                contamination=contamination,
14                novelty=True
15            ),
16            'one_class_svm': OneClassSVM(
17                kernel='rbf',
18                nu=contamination,
19                gamma='auto'
20            )
21        }
22        self.scaler = StandardScaler()
23        
24    def fit(self, X: np.ndarray) -> 'AnomalyDetectionEnsemble':
25        """Entrena todos los modelos del ensemble."""
26        X_scaled = self.scaler.fit_transform(X)
27        for model in self.models.values():
28            model.fit(X_scaled)
29        return self
30        
31    def predict_ensemble(self, X: np.ndarray) -> np.ndarray:
32        """Predicción por votación mayoritaria."""
33        X_scaled = self.scaler.transform(X)
34        predictions = []
35        for model in self.models.values():
36            pred = model.predict(X_scaled)
37            predictions.append((pred == -1).astype(int))
38        return (np.mean(predictions, axis=0) > 0.5).astype(int)