Combining feature selection methods (filter and wrapper)

Hello, I designed an integrated framework for feature selection which combines feature ranking techniques with sequential forward feature selection to find the optimal subset of the most informative features. I implemented this framework using MATLAB functions ( rankfeatures and sequentialfs ) on two microarray data (breast cancer and leukemia). The results were fluctuated between them, it was high accuracy for breast cancer and lower for leukemia. Is there any theoretical base behind this fluctuation in results.

Regards

Answer 1

[Walter Roberson]

If you implemented your code correctly and you get this result, then you have a theoretical research question, not a MATLAB question, and should be investigating through other resources.

If you implemented your code incorrectly then the error is on line 42, starting at column 9.

[Ahmed Alabdaly]

Dear Walter Roberson, Yes, the MATLAB code has worked properly. I do agree with you, I tried to find an academic resource related to bioinformatics. Thank you so much for your support. Today, I found the answer in "Bioinformatics: The Machine Learning Approach" second edition.

Best regards