python - How to normalize a confusion matrix?

Question

I calculated a confusion matrix for my classifier using the method confusion_matrix() from the sklearn package. The diagonal elements of the confusion matrix represent the number of points for which the predicted label is equal to the true label, while off-diagonal elements are those that are mislabeled by the classifier.

I would like to normalize my confusion matrix so that it contains only numbers between 0 and 1. I would like to read the percentage of correctly classified samples from the matrix.

I found several methods how to normalize a matrix (row and column normalization) but I don't know much about maths and am not sure if this is the correct approach. Can someone help please?

question from:https://stackoverflow.com/questions/65837460/how-to-normalize-confusion-matrix

Answer 1

I'm assuming that M[i,j] stands for Element of real class i was classified as j. If its the other way around you are going to need to transpose everything I say. I'm also going to use the following matrix for concrete examples:

1 2 3
4 5 6
7 8 9

There are essentially two things you can do:

Finding how each class has been classified

The first thing you can ask is what percentage of elements of real class i here classified as each class. To do so, we take a row fixing the i and divide each element by the sum of the elements in the row. In our example, objects from class 2 are classified as class 1 4 times, are classified correctly as class 2 5 times and are classified as class 3 6 times. To find the percentages we just divide everything by the sum 4 + 5 + 6 = 15

4/15 of the class 2 objects are classified as class 1
5/15 of the class 2 objects are classified as class 2
6/15 of the class 2 objects are classified as class 3

Finding what classes are responsible for each classification

The second thing you can do is to look at each result from your classifier and ask how many of those results originate from each real class. Its going to be similar to the other case but with columns instead of rows. In our example, our classifier returns "1" 1 time when the original class is 1, 4 times when the original class is 2 and 7 times when the original class is 3. To find the percentages we divide by the sum 1 + 4 + 7 = 12

1/12 of the objects classified as class 1 were from class 1
4/12 of the objects classified as class 1 were from class 2
7/12 of the objects classified as class 1 were from class 3

--

Of course, both the methods I gave only apply to single row column at a time and I'm not sure if it would be a good idea to actually modify your confusion matrix in this form. However, this should give the percentages you are looking for.