Summary

Let’s summarize what we have learned in this section.

• GLMs: We learned that GLMs are an extension of linear models (LMs) and are used to model non-normal data that fits several common error distributions. The coding is almost identical in both models, but GLMs include an argument to specify the error distribution family that’s going to be used.

• Error distribution: We discussed that it’s crucial to assess the diagnostic plots of the model residuals and the dispersion of the model to make an informed decision about which error distribution fits the data best and if it meets the model’s assumptions.

• Model selection and reduction: We also observed the model selection and reduction techniques to find the most appropriate combination of predictors and/or the minimal adequate model.