My company has a set of insurance claim data where I would like to compare against industry claim data. The rate is expected claim rate per month hence it is fairly small (around 0.X%). My preference is to use GLM to fit the model and find any factors that drive the difference between my own company's claim rate and industry rate.
My data sets include all policies with label 0 or 1 where 1 means claim and 0 means not claim. Would Poisson regression or Logistic regression be an appropriate choice in this case? And what would be the best tool to evaluate the performance of the model?