I am trying to properly display my confusion matrix for a multilabel classification. My approach to achieve that is following:
from sklearn.metrics import multilabel_confusion_matrix, accuracy_score
ypred = res
ytrue = np.argmax(y_test, axis=1).tolist()
ypred = np.argmax(ypred, axis=1).tolist()
multilabel_confusion_matrix(ytrue, ypred)
array([[[171, 0],
[ 0, 8]],
[[173, 1],
[ 0, 5]],
[[171, 0],
[ 0, 8]],
[[173, 0],
[ 1, 5]],
[[172, 0],
[ 0, 7]],
[[171, 0],
[ 0, 8]],
[[177, 0],
[ 0, 2]],
[[170, 0],
[ 0, 9]],
[[172, 0],
[ 0, 7]],
[[174, 0],
[ 0, 5]],
[[170, 1],
[ 0, 8]],
[[178, 0],
[ 0, 1]],
[[169, 2],
[ 0, 8]],
[[169, 1],
[ 1, 8]],
[[174, 0],
[ 0, 5]],
[[172, 0],
[ 1, 6]],
[[173, 0],
[ 0, 6]],
[[174, 0],
[ 1, 4]],
[[169, 0],
[ 0, 10]],
[[172, 0],
[ 0, 7]],
[[172, 1],
[ 1, 5]],
[[171, 0],
[ 0, 8]],
[[171, 0],
[ 1, 7]],
[[174, 0],
[ 0, 5]],
[[174, 0],
[ 0, 5]],
[[171, 0],
[ 0, 8]],
[[171, 0],
[ 0, 8]]], dtype=int64)
from sklearn.metrics import confusion_matrix
cm = confusion_matrix(ytrue,ypred)
It gives me the following output:
array([[ 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 6,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0,
0, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 10, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
0, 0, 0, 0, 5, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 8, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
0, 0, 0, 0, 0, 0, 7, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 5, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8]], dtype=int64)
When I try to display the confusion matrix as an image it gives me this output :
I wanted to label the classes of my confusion matrix so referred to the official site of scikit-learn and figured that they have a separate class named ConfusionMatrixDisplay() to help display a confusion matrix.
However, when I try to do it using the ConfusionMatrixDisplay, I try out the following code:
import numpy as np
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay
labels = actions
fig, ax = plt.subplots(figsize=(9, 9))
ConfusionMatrixDisplay.from_predictions(
y_true,
y_pred,
display_labels = labels,
xticks_rotation=45,
ax = ax,
colorbar = False)
But this time it is showing a different type of confusion matrix:
Notice the values are not diagonal but scattered. How can I get the same result as the first image (using seaborn) for this approach using ConfusionMatrixDisplay?
How can I put the X-axis labels on the top for the ConfusionMatrixDisplay approach? How can I edit the "True Label" and the "Predicted label" in this case?
I also want to know how I can put the X-axis labels on the top for the seaborn approach and rotate them in 45 degrees so that they do not look overlapped.
