phik_correlation_matrix_pandas_car_insurance_dataset.py

1import pandas as pd
2import phik
3from phik import resources
4from phik.report import plot_correlation_matrix
5
6# Load the car insurance dataset
7data = pd.read_csv(resources.get_resource('car_insurance_prediction.csv'))
8
9# Calculate the phik correlation matrix between all variables in the dataframe
10phik_overview = data.phik_matrix()
11
12# Plot the correlation matrix
13plot_correlation_matrix(phik_overview.values, 
14                        x_labels=phik_overview.columns, 
15                        y_labels=phik_overview.index, 
16                        vmin=0, vmax=1, color_map="Greens", 
17                        title=r"$\phi_K$ correlation", 
18                        fontsize_factor=1.5, 
19                        figsize=(10, 8))