This dataset is of an experiment using the internal combusion engine (like what cars have). We are trying to predict a pollutant \(y=\) NOx (the concentration of nitrogen oxides in the exhaust) based on a measure of how rich or lean the fuel–air mixture is, \(x=\) E. Slide the degree and watch the two losses: training loss can only fall, validation loss falls and then rises, and the gap between them is what overfitting looks like. The loss axis is fixed at 0–1, so once a loss is off the chart it stays off the chart (the readouts still give it). By degree 43 the polynomial passes through all 44 training runs exactly — training loss 0 — and shoots off between them; the validation loss is then around \(10^{22}\).