I am trying to normalize my features for a classification model with 3 class outputs. There are two kinds of features. First is medical test results and second is patient information such as age. The first kind of features is normalized with given ranges(0 is the lower normal value and 1 is the upper normal value. anything which is less than 0 or greater than 1 is considered abnormal). For obvious reasons, I don't have such ranges for age and hence I am using mean and standard deviation for normalizing age.

The problem is, the two kind of features do not follow same kind of distribution. While the first kind of features have little density outside 0 or 1, the second has greater density.

I am wondering if this will make the second kind of parameters more important (age) in the svm classification as it will contribute to the distance more than the first kind of features

The first feature distribution k

The age distribution(added 0.5 to have consistance centres) enter image description here


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.