SVM and de-normalization, should it be done?
I have an SVM model for training , and a sample data which are used at an "apply model" block . The main goal is to predict total month sales.
I normalize both data with a "Normalize" block.
The problem is that after normalizing both model and data, I end up having results on the normalized scale.
My question is, what can i do to de-normalize/transform the values in the original sales range?
I am using a normalization between 0 and 1.
The prediction is being made for months where i know the sales values for testing purpose.
|OriginalSales||Normalized Sales||Normalize Prediction||OriginalScalePrediction|