Hi! Could you explain more about what would be the advantages of using Normalization instead of Standardization?
There is a general acceptance that if your data varies a lot, you should use standardization.
That’s because normalization bounds the data within [0,1] which automatically means that you lose a lot of information about outliers.
Standardization on the other hand does not bound the data. It rescales it to have a 0 mean and 1 standar deviation, so most data falls between -3 to +3, but you can still spot outliers (e.g. numbers like 4.5 would be outlers).
Some models may fair better under standardizaiton, while others – under normalization. For instance in clustering, you would prefer standardizaiton, because losing information about outliers will greatly alter the clustering solution.
The 365 Team