Employing the kNN model in a regression problem

Although used predominantly to solve classification problems, the k-Nearest Neighbors model that we saw in Chapter 3, Classification Techniques, can also be used in regression models. This recipe will teach you how it can be applied.

Getting ready

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

How to do it…

Again, using Scikit to estimate this model is extremely simple (the regression_knn.py file):

import sklearn.neighbors as nb

@hlp.timeit
def regression_kNN(x,y):
    '''
        Build the kNN classifier
    '''
    # create the classifier object
    knn = nb.KNeighborsRegressor(n_neighbors=80, 
        algorithm='kd_tree', n_jobs=-1)

    # fit the data
    knn.fit(x,y)

    # return the classifier
    return knn

How it works…

First, we read the data in and split it into the dependent variable y and independent variables x_sig; we are selecting only the significant variables that we found earlier,...