Here are my comments on the final milestone. Overall, really excellent work, this are mainly things to keep in mind for the future:
Clearly state population of interest and sample
Was the age distribution your only important finding from the EDA?
Very good discussion of assumptions
- There are discrepancies between the values you report in text and those shown in the outputs. Most are minor but it’s still a little confusing.
Give the units for the coefficient (it’s the probability for linear models, log odds for ordinal)
Remember that you can’t interpret AIC or likelihood in isolation. Only relative to other models (as you did later on).
Be careful with the Bayesian vs. Frequentist interpretations of confidence intervals (this is frequentist)
Don’t interpret those intercept cutpoints- that’s how the model is converting a continuous variable back to a categorical variable under the hood
You didn’t test the proportional odds assumption! It’s a hypothesis test or visualization in R. For example: this package
Here are my comments on the final milestone. Overall, really excellent work, this are mainly things to keep in mind for the future: