0

I have an input parameter that have 200 values. However, among the 200 values, there are only 3 distinct values. For example, like this:

X1

10 14 14 10 22 22 10 10 14 . . .

Should I treat this parameter as a categorical input and encode it before feeding to a neural network model?

However, if the parameter is treated as a categorical input, then it cannot accept any new values? So do we still consider it as numerical input?

Steven Chan
  • 1
  • Give us some context about this variable, what it represents, and why you only have 3 distinct values. – user2974951 Nov 22 '22 at 09:57
  • The parameter from a manufacturing process. Each value is the average value of that parameter X1 during the processing of a batch. If there are 10 values for X1, then there are 10 batches.

    Hope this clarifies the context

     – Steven Chan Nov 22 '22 at 15:01
  • If this is a parameter which can take on different values in the future then it might be better to code is as numerical... although if the new value will be quite different from the ones that the model has seen the results might be less than ideal. – user2974951 Nov 25 '22 at 12:09

0 Answers0