Hi Raaid,

Please be more specific, what you mean by filtering a large data set?

Do you want to get rid of noisy, illegitimate values? or

Do you want to select best subset of features? or ....

And please provide some info about your data as well.

Regards

Ahmed

Thank you so much; I mean here how to select best subset of features

Hi,

This mainly depends on your task: wrapper/filter--based on that you need to use a search or rank method, respectively--to find the selected feature subset. Then, utilize an evaluation algorithm to evaluate the merit of the selected subset. All this process has to be done under the evaluation of, e.g., cross-validation.

HTH.

Dr. Samer Sarsam

Thank you, dear Dr. Samer, for answering. My idea is like this: I am planning to design a new mechanism that has an ability to detect untrustworthy Interest packet incoming to the PIT, which is generated by the attacker.

Packet(Internet) sniffer algorithm needs to be looked into . It is network layer +Data link at least .

Dear Raaid Alubady,

In this case, feature selection can be applied in the following ways:

- either the important features subset can be selected as a part of the mechanism of the algorithm that you are developing. In this sense, selecting the features is always necessary as one of the algorithm. This has to be accomplished after cleaning the data up, or

- it has to be done as one of the preprocessing steps, i.e., before building the predictive model. Hence, feature selection here is optional.

Regards,

Dr. Samer Sarsam

Thank you so much Dr. Samer

Thank you so much Aparna Sathya Murthy

Dear Raaid see this paper is using the ML for filtration process,

www.ijarcst.com/doc/vol2-issue2/gparvathi.pdf

regards