E4. To be completed after lesson 14¶
Why are prior predictive checks useful and important?
I contend that plotting the “best fit line” is not a sufficient graphical model assessment. I further contend that graphical posterior predictive checks are better. Do you agree? Why or why not?
If a model’s posterior predictive checks check out, does it mean the model is a good generative model?
Did you read the solutions to homework 3?
Write down any questions you may have.