Classification Learner and sequentialfs

2 views (last 30 days)
I use the Classification Learner to select the prediction model that best classifies my data. How can I take advantage of Matlab's sequentialfs to select the best possible features for my data? I tried to export the model, or to export the code generated by the Learner, and then combine the model or the code with sequentiafs with no success so far. BTW, it would be great to add an automatic feature selection option in the next version of the Classification Learner. The manual feature selection is helpful, but sequentialfs seems to automate the whole process, which helps a lot when the number of features is high. Thanks

Answers (1)

Sayan Saha
Sayan Saha on 11 May 2018
Here is an example showing how to use "sequentialfs" with "fitglm" for fitting a logistic model to the data set. Similar steps can be followed for classification with "sequentialfs" as well. You will be performing the classification within the "critfun" function and returning the measure indicating the performance of the classifier.
  1 Comment
Giovanni Barbarossa
Giovanni Barbarossa on 15 May 2018
Thank you. I found plenty of examples. This blog is also very useful https://blogs.mathworks.com/loren/2011/11/21/subset-selection-and-regularization/

Sign in to comment.

Community Treasure Hunt

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

Start Hunting!