How to build a regression model with a continuous response variable bounded from 0 to 1?

I think it is not logistic regression, where I am not predicting a binary response variable, Right?

Sorry for duplication if any, I tried to search but not find. (I feel this question must be asked many times.)

  • $\begingroup$ It's called a beta regression. I believe your data can't include 0 or 1 for this to work properly though. $\endgroup$ – Michael Webb Sep 26 '17 at 15:38
  • $\begingroup$ @Great38 can we do some thing like $\text{min}~ \|\text{sigmoid}(\beta^Tx)-y\|^2$ $\endgroup$ – Haitao Du Sep 26 '17 at 15:39
  • $\begingroup$ @MatthewDrury thanks for helping me to find it. One more question, can we do $\text{min}~ \|\text{sigmoid}(\beta^Tx)-y\|^2$? $\endgroup$ – Haitao Du Sep 26 '17 at 15:42
  • $\begingroup$ I'm not sure, I haven't actually every used beta regression but I've seen a lot of resources on this site. $\endgroup$ – Michael Webb Sep 26 '17 at 15:45

Browse other questions tagged or ask your own question.