I have a dataset with 134 attributes and my goal is to build a binary classification model. While exploring the dataset, I found that there was high skewness present in the attributes. I wanted to understand that would the high skewness present in the attributes have any impact on the classification model I'll build.

For example, if there are attributes A,B,C,D,E,F and Y (response variable) and A, C, E and D seem to have high skewness (both towards positive and negative sides) then do the labels which my model generates for response variable Y get impacted because of this skewness.

Please note that I know how to handle class imbalance, and this is not a class imbalance problem. I also know that skewness can be removed using boxcox transformation from the scipy package , but wanted to be sure if this data needs it or not, because from what I've read, skewness has an impact on regression models, and I couldn't find anything about classification.

  • $\begingroup$ If you have skewness in both the lower and upper bound than no transformation is going to help there. Also is this data coming from questionnaires (or related to)? Also you mention classification models and then regression? So which is it? Regression does not have a problem with skewed data. $\endgroup$ – user2974951 Feb 19 '19 at 7:53
  • $\begingroup$ This question is not from a questionnaire, I am working on this dataset and wanted to know if skewness present would affect my classification model results.Also, my research about skewness told me that it does affect a regression model. This was a classification problem. $\endgroup$ – Shekhar Tanwar Feb 20 '19 at 4:39
  • $\begingroup$ hey did u find an answer? $\endgroup$ – Srinath Ganesh Oct 10 '20 at 17:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.