Does any one knows hoe to perform the average importance of the input variables in predicting the output in Matlab?
I am appreciate if share your idea.
It depends on the experimental results.
Each time you can fix all the inputs except one, and then you can differentiate between the obtained results.
Analyzing the correlation between different variables in the input data, can help in identifying the importance of variables and can also help in improving the output. You may refer to following links:
http://in.mathworks.com/help/matlab/ref/corrcoef.html#responsive_offcanvas
http://in.mathworks.com/matlabcentral/fileexchange/56937-feature-selection-library
Hi Maryam,
As Farhan said, You keep change one variable at a time and notice the difference in output. Do the process for other variables. Do not apply it for the whole program and take out the part of a program whichever is necessary and run the program. After identifying the importance of particular variable, keep change the value of the variable from wide to narrow range. Hope it would be helpful.
Thanks & Regards,
M.N.Sumaiya
Fron all info sources i see most useful query "importance variables Matlab" in google
Be happy
You can also check if this is of any help to you.
http://www.palisade.com/downloads/pdf/academic/DTSpaper110915.pdf
Please kindly find the answer in following link:
https://www.mathworks.com/matlabcentral/answers/299646-how-to-obtain-the-relative-importance-of-each-input-variable-for-a-neural-network#answer_783049
- Badges
- Science topic
- Similar topics
- Filing