Hello,
I am coming from insurance business. Im building glm models using commercial software.
When i fit a new variable I look into fitted vs observed plots, relativities plots (i think this is only relevant in insurance business), check if my new variable significantly reduces deviance. Once you think you are close to finish modelling - look into deviance residuals. Basically, i have refined my model building process using commercial software.
I want to diverge perhaps from insurance business building models in R. Please, could you share your best practice when building a glm model in R?
Thank you