R: Linear or GLM: weighted results: wls:

So probably a basic enough question

What i want to achieve is to do a regression of lapse rates on savings type policies. So say person a in one yr withdraws 10 out of 100 and person b 10 out of 50. So the weighted lapse rate is 13.3% and not 15% (both weighted equally)

I have thousands of data points with many zero instances.

1. My understanding of glm in R is that this is better for count data and not weighted result data? (I would get the 15% result)
2. If i group the data would the results be useful? Say if i put it thru a zero inflated poisson model?
3. Should i instread just be using linear regression with a number of different possible models?

Many thanks K