AnomalyClassifier.py

import numpy as np
from sklearn.base import BaseEstimator, ClassifierMixin

from sklearn.ensemble import IsolationForest
from sklearn.covariance import EllipticEnvelope
from sklearn.svm import OneClassSVM


def mode(x):
    (values, counts) = np.unique(x, return_counts = True)
    idx = np.argmax(counts)
    return values[idx]


class AnomalyClassifier(BaseEstimator, ClassifierMixin):
  
    '''
    Use estimator to classify the majority class in y as inliers
    and the minority class as outliers. If target variable
    y is not used, use X dataset to fit the anomaly detection
    estimator.
    '''
    
    def __init__(self, estimator):
        self.estimator = estimator
    
    def fit(self, X, y = None):
        if y is not None:
            idx = y == mode(y)
            X = X[idx, :].copy()
            y = y[idx].copy()
        return self.estimator.fit(X, y)
    
    def predict(self, X):
        return (self.estimator.predict(X) == -1).astype(int)