gam.py

from sklearn.ensemble import RandomForestRegressor
from sklearn.base import clone
import numpy as np

class GAM(object):
    """
    Simple generalized additive model which fits each feature w.r.t
    the output independently using a model specified by
    base_estimator. the final predictions can be done by regressing linearly
    the predictions of the model of each features by using a pipeline.
    Example:
    
    from sklearn.pipeline import make_pipeline
    from sklearn.linear_model import LinearRegression
    
    reg = make_pipeline(GAM(), LinearRegression())
    reg.fit(X, y)
    
    """
    def __init__(self, base_estimator=RandomForestRegressor()):
        self.base_estimator = base_estimator
        self.estimators_ = None
        self.agg = None

    def fit(self, X, y=None):
        self.estimators_ = [clone(self.base_estimator) 
                            for i in range(X.shape[1])]
        for i in range(X.shape[1]):
            self.estimators_[i].fit(X[:, i:i+1], y)
        return self

    def transform(self, X):
        y_pred = np.empty_like(X)
        for i in range(X.shape[1]):
            y_pred[:, i] = self.estimators_[i].predict(X[:, i:i+1])
        return y_pred

    def predict(self, X):
        return self.transform(X).mean(axis=1)