Supervised machine learning

[lgatto]

• The supervised machine learning should start with and intro (kNN is good), follow up with in-sample and out-of-sample, move to caret::train, and the follow up with caret and not revert back to glm before introducing confusion matrices.

• The main exercise about model selection should come as soon as possible.

[FSuessenbach]

Thanks for this great course! I spotted a minor error you might want to fix. In the SLM Cross-validation section (5.3.2) you forgot some brackets in the calculation of the RMSE for the Boston data set challenge.

Your code reads: sqrt(mean(p - Boston$medv)^2) #6 6.647482e-14

However it should be: sqrt(mean(((p - Boston$medv)^2))) #4.679191

[lgatto]

Indeed, thanks!