Is the predictor function in logistic regression a function of the dot product of the parameter vector and the feature vector onto a real value between zero and one? If I'm getting dot products outside of that interval, does that mean my parameter vector is bad?



In Logistic regression, the result of the linear predictor (g(x)) is from negative infinite to positive infinite.

enter image description here

Then the result of (pi(x)) ranges from 0 to 1 via the link function. This is the definition.

  • $\begingroup$ Great, thanks! For future readers, once you've done the dot product to get a number (which is really the Y coordinate in linear space... the answer to a linear regression problem in terms of log-odds) you transform it back into probability space by doing the following (where p is your answer): exp(p) / (1 + exp(p) ) . The resulting answer should be between 0 and 1. $\endgroup$ – Walrus the Cat Sep 5 '13 at 9:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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