[Python] Do I need to use scaler even if my dataframe has fairly normalized data within a...

I am trying to perform LSTM on a Training dataset with below structure for forecasting



There are about 2 million such sets in the dataframe. So the shape will be something like (2M, 10 or 15, 3). It could be lesser if I categorize the data on other factors
Column A & B always start with 1 and then change (Generally 0-2 % either side, rarely 10%, never 20%). Column C is always between 0 and 100.

My question is : Do I still need to use minmax or Robust Scaler, even though my data is already fairly normalized within a range. Also if I should still use it, should the range be (0,1) or (-1,1)

Continue reading...