What is the best way to scale a numerical dataset [closed]

Question

I have a dataset with differents attributes which don't have the same range on their values which is a problem when we need to compute distance beetween objects. After some research i found that i can do the regularisation job with this formula : (value-min)/(max-min) where min and max are respectively the minimum and maximum value in the domain of val attribute. The question is that one, does it exist other ways ?

Thank you for your help.

Answer 1

There is pretty much mess in terminology in your question :). Data Regularization is used for model selection, it is not about data processing. Here it is described in more friendly manner.

What you mean is Feature Scaling. It can be done in several ways including Rescaling, the method you described. You may also use Standardization (normalization) and Scaling to unit length.

These answers may be helpful:

• Normalization vs Scaling
• Normalization vs Standardization