you may follow these below papers,

http://keddiyan.com/files/AHCI/week2/8.pdf

http://pdgindia.com/uploads/pdf_(published_version)28620167829.pdf

http://www.nsclab.org/yang/publications/TPDS-2011-10-0744.pdf

Thanks Prof. Chakraborty. I have gone through all the materials. I have read them all before now. .

Hi,

I suggest the read of the follow paper:

https://www.researchgate.net/publication/285403911_Social-Spider_Optimization-based_Support_Vector_Machines_applied_for_energy_theft_detection

Article Social-Spider Optimization-based Support Vector Machines app...

Thank you Prof. Danillo. I believe the material will be very useful.

You may try techniques from the field of deep learning because deep neural networks learn in theory useful features for the respective prediction or classification task by themselves ("feature learning").

Thank you Prof.Danillo.. It's very interesting paper.. 

I recommended use Self Organizing Maps, Principals Components Analysis or Undiscriminating Linear Discriminant Analysis as Featuring Selection Method to try optimize network traffic analysis.   

I am not expert in traffic studies, but regarding the feature selection technique it would help you to find the most relevant features for your problem, simplifying the interpretation of the problem and spedding up the learning process.