Machine learning methods that hold up

Steps

1. 1

   Step 1 · 3 min read

   Train, validation and test sets explained

   The foundation of honest evaluation.

2. 2

   Step 2 · 3 min read

   Overfitting and underfitting in machine learning

   Read what the train-validation gap is telling you.

3. 3

   Step 3 · Lab, 30 minutes

   Measure overfitting with validation curves

   See overfitting appear on a chart.

4. 4

   Step 4 · 3 min read

   Cross-validation explained: k-fold, stratified and grouped

   Stable estimates from limited data.

5. 5

   Step 5 · 3 min read

   Precision, recall and F1 score explained with an example

   Accuracy lies on imbalanced data.

6. 6

   Step 6 · 3 min read

   Data leakage in machine learning: how good scores lie

   The most common reason good scores fail.