Hi all,

I have a continuous dependent variable varying between 0 and 1. There are lots of zeros, too:
Array

When I run a regular OLS, I receive negative fitted values. So I decided to run a GLM to make sure the fitted values are in the right place. I've read lots of things about the selection of family and link selection but I still don't understand the mechanism behind it. Residuals from the OLS are normally distributed. Can I use Gaussian family and log link? This combination gives me all positive predicted values.

Many thanks.