Estimating the correlation between two variables with a contingency table and a chi-squared test

Whereas univariate methods deal with single-variable observations, multivariate methods consider observations with several features. Multivariate datasets allow the study of relations between variables, more particularly their correlation, or lack thereof (that is, independence).

In this recipe, we will take a look at the same tennis dataset as in the first recipe of this chapter. Following a frequentist approach, we will estimate the correlation between the number of aces and the proportion of points won by a tennis player.

How to do it...

1. Let's import NumPy, pandas, SciPy.stats, and Matplotlib:

   >>> import numpy as np
       import pandas as pd
       import scipy.stats as st
       import matplotlib.pyplot as plt
       %matplotlib inline

2. We download and load the dataset:

   >>> player = 'Roger Federer'
       df = pd.read_csv('https://github.com/ipython-books/'
                        'cookbook-2nd...