I have a dataset with different types of numerical values (both negative and positive numerical values) for the inputs (for example, -40, -35, 1, 25, 39, etc., that is, multiple inputs) and single output numerical value (either negative or positive).

I have tried to use linear regression, but I haven't been so successful and I think one of the reasons is negative values.

What is the best way to deal with this scenario? What model should I use?

I am using Keras for my AI model.

  • 1
    $\begingroup$ Can you provide more details? e.g. how many input features do you have? Are they all integer? Output value is limited to a range or could be any integer? $\endgroup$
    – aminrd
    Dec 5, 2019 at 20:59
  • 2
    $\begingroup$ negative values are not the problem, you need to scale your data. For example if your data range is $[-200, 200]$ then scale the data to range $[-1, 1]$ or you could scale them to have mean of $0$ and standard deviation of $1$. You can use scikit-learn library for data preprocessing for example. $\endgroup$
    – Brale
    Dec 5, 2019 at 21:46


You must log in to answer this question.