0
$\begingroup$

I am new in statistical modeling and l am trying to model a financial data with logistic regression and I am facing a challenge here.

  1. Some of the predictor variables have integer values( negative, zero and positive) in them. How to I treat them? In school professors dont use negative data values in teaching.

  2. The dependent variable(binary) has about 97% of zeros and 3% of ones and I am modeling the case where the dependent variable is 1. Since the zeros are so much, do I take a reasonable sample (like 2/3)of data so that the ratio of the ones to zeros will be say 1:3?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ 2. See here. The only reason to do that would be constraints on computational resources. $\endgroup$
    – Scortchi - Reinstate Monica ♦
    Commented Mar 10, 2014 at 12:47

1 Answer 1

Reset to default
3
$\begingroup$

Welcome to the site.

Regarding 1) It shouldn't matter if the predictor variables are negative or positive.

Regarding 2) The key thing is usually whether the sample of DV = 1 is large enough, not whether it is much smaller than that of DV = 0.

Improve this answer
$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.