I want to calculate sensitivity, specificity, PPV, and NPV from a confusion matrix for multiclass classification. My objective is to learn the basic concept of constructing a confusion matrix for multiclass classification and how to calculate sensitivity and specificity from a confusion matrix manually without using machine learning or other programs.
My Confusion matrix:

First, I wanted to know whether my confusion matrix looks okay
Second, how can I calculate sensitivity, specificity, PPV, and NPV for different levels for the predicted class without using any program?
Third, I want to learn what's the best way to present sensitivity and specificity in a confusion matrix.