Can standardisation and outlier treatment be done on discrete numerical feature?

Question

I have a column in my dataset which has the number of payments made to the physicians. This is a discrete nunerical feature. I know that transformations and scaling can be done for continuous data. But what about discrete numerical features?

When I plot a histogram it gives me frequency distribution. Looking at the histogram standardisation, outlier treatment do not make sense. Can someone tell me how to deal with discrete numerical data. There are a lot of articles which talk about continuous data as well as categorical but I couldn't find proper articles which tell how to deal with disrete data

Answer 1

You can definitely do standardization and outlier treatment for discrete numerical feature. But the choice of doing it or not depends on the use case.

For example -

1. Training a decision tree model do not require features to be standardized.
2. Training linear models with regularization requires all features to be in similar range.
3. For neural networks standardization helps in reaching the convergence faster.