search

InitRoot / NSLKDD-Dataset

NSLKDD Dataset for WEKA
MIT License
35 stars 24 forks source link

Attribute selection #5

Open aalces opened 1 month ago

aalces commented 1 month ago

Hello! I'm currently working with my university project using this dataset. Could you please tell me if the label (xAttack) in the test set can affect the final result as it can be used as an attribute for classification? Or I probably misunderstand something. Another question is about attribute selection via PCA+Ranker with default settings. For some reason I get 79 attributes as the result. Could you please tell me why this could be happening and how should i fix this?

Thank you in advance.

InitRoot commented 1 month ago

Hi,

Yes, any features/attributes added to a dataset when learning will impact the results. xAttack, can be used to split the dataset into sub datasets training per attack type, if that makes sense.

I've not been involved with feature selection for a number of years, so I won't be able to assist with it. General rule of thumb, ensure you understand exactly how your attribute selection method works.