CODEWITHSUNDEEP
datasetnormalizationdata-mining
Musa Bilal
Musa Bilal

Reputation: 1

While carrying out data normalisation should I apply different techniques on different features

I have been trying to normalize my data set which contains a number of features with different distributions, normalized, positively skewed and negatively skewed. While carrying out data normalization as a preliminary step to building up a classification model, should I apply the same normalization technique to all features within the data set or should I apply different normalization techniques for different features within the data set?

Upvotes: 0

Views: 239

Answers (1)

Has QUIT--Anony-Mousse
Has QUIT--Anony-Mousse

Reputation: 77485

Maybe.

There is no one-size-fits-all rule.

If you e.g. have x,y coordinates on a small scale, yes then using the same normalization is probably more appropriate. Same as in "same scale", not just "same method". If you have latitude and longitude, do not do any normalization, unless you have reasons to assume earth was distorted. And if you have completely different scales, say attribute A is "shoe-size", and attribute B is "pi / shoe-size", then YES, you will need different normalization for these two attributes.

Usually, when this question arises, you have not yet understood your data.

Upvotes: 0

Related Questions