Hello, everybody, I would like to ask you the following hesitations:

I have two data frame (training and test data) (A, B) and four situations:

1) First scenario: First, I calculate the feature extraction process from my input data (A, B), and then I do standardization or Normalization of this process. This with the data combined in a data frame C.

2) Second scenario: First, I calculate the standardization or Normalization of my inputs data (A, B) then perform feature extraction. This with the data combined in a data frame C.

3) Third scenario: First, I calculate the feature extraction process from the input data (A, B), and then I do standardization or Normalization of this process, respectively. I mean taking them as two different data frames.

4) Fourth scenario: First, I calculate the standardization or Normalization of my inputs data (A, B), respectively (I mean taking them as two different data frames), then perform feature extraction.

Depending on which scenario I choose, how can it affect the accuracy of the algorithms for classification or detections?

It’s known that Normalization typically means rescales the values into a range of [0,1]. Standardization typically means rescales data to have a mean of 0 and a standard deviation of 1 (unit variance)

The input data have different data ranges, such as 0 to 5, 0 to 0.015, or 0 to 2.

How can the algorithm’s performance affect whether or not the input data are analyzed separately or combined?