Hi Rishika,
As per my understanding, you want to calculate the accuracy of an SVM model for multiple classes whose labels are predicted based on one-vs-all (also known as one-vs-rest) approach.
If you have trained the SVM model based on one-vs-all approach and have the test data and corresponding true labels. Then make predictions for the test data using your approach. By comparing the predicted labels (predictions) with the true labels, you can calculate the accuracy by dividing the number of correctly predicted data points by the total number of data points.
Refer to the below documentation to learn more about “Support Vector Machine Classification” using MATLAB.
