Estimating the output of an electric plant using CART

Previously, we used decision trees to classify our bank contact calls (refer to the Classifying calls with decision trees recipe from Chapter 3, Classification Techniques). Classification and regression trees are the equivalent methods applied to regression problems.

Getting ready

To execute this recipe, you need pandas and Scikit. No other prerequisites are required.

How to do it…

Estimating CART with Scikit is extremely easy (the regression_cart.py file):

import sklearn.tree as sk

@hlp.timeit
def regression_cart(x,y):
    '''
        Estimate a CART regressor
    '''
    # create the regressor object
    cart = sk.DecisionTreeRegressor(min_samples_split=80,
        max_features="auto", random_state=666666, 
        max_depth=5)

    # estimate the model
    cart.fit(x,y)

    # return the object
    return cart

How it works…

As with all the other recipes, we first load the data and extract the dependent variable y and independent variables...