Hi Niloufar,

Both WEKA and Matlab are fine.

If you are using WEKA, you could deal with one of the following:

- Applying the "CfsSubsetEval" in conjunction with the "GreedyStepwise" (for example). (CFS uses a heuristic to score feature subsets that trades off the average correlation to the class against the average intercorrelation. Broadly speaking, feature subsets with high average correlation to the class and low intercorrelation receive higher scores. If two features are perfectly correlated, then only one will make it into the final subset. Which one depends on the search strategy (i.e. which of the two is considered first by the search strategy.)

- Applying "CorrelationAttributeEval" in conjunction with a "Ranker" method. CorrelationAttributeEval evaluates the worth of an attribute by measuring the correlation (Pearson's) between it and the class.

- You can apply the "LinearRegression" algorithm that performs by default feature selection. In this context, you can bump up the ridge parameter to perform heavier regularization if the correlation causes problems-- you should turn off elimination of colinear attributes and attribute selection in LinearRegression as well: those heuristics aren't normally necessary when performing proper regularization.

HTH.

Samer Sarsam, PhD.

Thanks for your complete answer dear Samer. By the way, because I am new in working with Weka, I am a little confused to deal with these solutions. So, is there any source code or website that you suggest to make it more clear for me?

Regards,

Niloufar

No worries, Niloufar.

The above approaches can be applied through Weka GUI Chooser (the "Classify" tab in the "Explorer", in particular). No code is required at this interface.

Now, if you need to write some code, then you could look at the following links:

https://weka.wikispaces.com/Primer

http://weka.sourceforge.net/doc.dev/weka/attributeSelection/CfsSubsetEval.html

http://weka.sourceforge.net/doc.dev/weka/attributeSelection/GreedyStepwise.html

http://weka.sourceforge.net/doc.dev/weka/attributeSelection/CorrelationAttributeEval.html

http://weka.sourceforge.net/doc.dev/weka/attributeSelection/Ranker.html

Cheers,

Samer Sarsam, PhD.

Super thanks.