|
def get_most_correlated_variables(corr, num_pairs=10): |
|
correlation_melted = pd.melt(corr.reset_index().rename(columns={"index": "var_1"}), id_vars=("var_1"),var_name='var_2') |
|
correlation_melted = correlation_melted[correlation_melted.var_1!=correlation_melted.var_2] |
|
correlation_melted['var_couple'] = correlation_melted[['var_1','var_2']].apply(lambda x:tuple(sorted([x[0],x[1]])), axis=1) |
|
correlation_melted = correlation_melted.drop_duplicates(subset='var_couple').drop(['var_couple'],axis=1) |
|
correlation_melted['abs_value'] = correlation_melted['value'].abs().round(3) |
|
return correlation_melted.sort_values(by='abs_value').tail(num_pairs).drop('abs_value', axis=1).reset_index(drop=True) |
|
|
|
def plot_correlation_matrix(X, features2): |
|
corr = X[features2].corr() |
|
# return the most correlated variables |
|
most_correlated_variables = get_most_correlated_variables(corr, num_pairs=10) |
|
max_correlation = 1.25*most_correlated_variables['value'].abs().max() |
|
# Generate a mask for the upper triangle |
|
mask = np.zeros_like(corr, dtype=np.bool) |
|
mask[np.triu_indices_from(mask)] = True |
|
# Set up the matplotlib figure |
|
f, ax = plt.subplots(figsize=(11, 9)) |
|
# Generate a custom diverging colormap |
|
cmap = sns.diverging_palette(220, 10, as_cmap=True) |
|
ax.set_yticklabels(features2, fontsize=18) |
|
ax.set_xticklabels(features2, rotation='vertical', fontsize=18) |
|
# Draw the heatmap with the mask and correct aspect ratio |
|
sns.heatmap(corr, mask=mask, cmap=cmap, vmax=max_correlation, center=0, |
|
square=True, linewidths=.5, cbar_kws={"shrink": .8}) |
|
return most_correlated_variables |
