MATLAB Answers

How do I find or visualise the top features used for SVM classifier?

10 views (last 30 days)
Soyoung Choi
Soyoung Choi on 14 Jul 2020
Answered: Aditya Patil on 17 Aug 2020
Hi, I am very new to matlab.
I am using 25 predictors to classify 5 different cell morphology types.
I used the Classification Learner app and the linear SVM to successfully classify to a good accuracy.
However, how would i look at the best features used to classify these morphologies? Especially once the trained model has been saved (i'm not sure if this makes a difference but have learned everything can make a difference with these things!)
I have tried looking at ClassificationECOC and can see there is 'learner weights' in the properties. Is there a way of 'opening' or visualising this?
Thank you in advance

  0 Comments

Sign in to comment.

Answers (1)

Aditya Patil
Aditya Patil on 17 Aug 2020
It's not possible to determine feature importance in SVM model, unless you use linear kernel. You can read more about this in this answer.

  0 Comments

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!