"Effective testing for machine learning systems requires both a traditional software testing suite (for model development infrastructure) and a model testing suite (for trained models)."
Read it!Prediction intervals are commonly used for linear models but are often underused for random forests. Leveraging the fact that a random forest can provide a conditional distribution instead of just the conditional mean makes prediction intervals relatively straightforward to use in this context.
Read it!"The difficulty is that machine learning is a fundamentally hard debugging problem. Debugging for machine learning happens in two cases: 1) your algorithm doesn't work or 2) your algorithm doesn't work well enough."
Read it!