How to modify Scikit-Learn's ConfusionMatrix low-level attributes
Today I learned how to modify low-level components of a ConfusionMatrixDisplay object, i.e. fiddling directly with fig and ax.
I needed to do this after trying to plot a gigantic matrix of nearly 100 labels.
I searched the documentation for ways to change the figsize value, but found out that the ConfusionMatrixDisplay implementation does not allow custom plot parameters to be passed.
This is the workaround suggested on Stack Overflow:
import numpy as np
from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix
import matplotlib.pyplot as plt
cm = confusion_matrix(np.arange(25), np.arange(25))
cmp = ConfusionMatrixDisplay(cm, display_labels=np.arange(25))
fig, ax = plt.subplots(figsize=(10,10))
cmp.plot(ax=ax)