1
$\begingroup$

I have 11 features with categorical values [1and0] and 4 features with price in dollars. So I was thinking to scale them.

Which type of scaler should be used?

1) MinMaxScaler 2) StandardScaler 3) Robust Scaler or 4) MaxAbsScaler

How can we know what type of scaler we should use?

Does using different scaler reduce my classification model performance?

$\endgroup$
0
$\begingroup$

There is no one way to drill down the exact scaler that you need to use. The best way is to look at the distribution of data by plotting Histograms, boxplots. You will understand how similar is it to the normal or gaussian distributions. That and some knowledge of the math behind the scalers will give you a great idea. For example if it looks similar to left skewed normal disribution then we can use something like log transformation. What you could do is use different tests to understand which distributions it most resembles to like KS Test (Kolmogorov-Smirnov Test) Again there is no 100% right solution for it. You need to use your expertise in stats to make the most suitable choice for your data.

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.