Notes from teaching

[amueller]

• Weird transition in notebook 1 after Iris?

• Not good examples for supervised and unsupervised learning in nb1.

• Second notebook: why homedir ?

• datasets: you can load anything into a numpy array.

• Why so much talk up front?

• explain Train_test_split fist without arguments

• Regression before classification? Neighbors before logistic?

• Mse for regression in linear regression Formular but then r2

• add precision-recall curve to roc curve stuff

• add binary classification confusion matrix before roc curve

• trees: max leaf nodes

• pca explained variance ratio

• cross-validation reimplementation is overkill?