Issue in notebook lecture 3

[nickvdw]

This is regarding 03 - Model Selection.ipynb.

The cell directly under

Model selection

• Again, we can compare multiple models by looking at the ROC curves
• We can calibrate the threshold depending on whether we need high recall or low FPR
• We can select between algorithms (or hyperparameters) depending on the involved costs.

that plots the ROC curves for SVC and RF together with the thresholds (associated cell is In [39]: in the currently uploaded notebook) used

plt.plot(fpr_rf[close_default_rf], tpr[close_default_rf], '^', markersize=10,
         label="threshold 0.5 RF", fillstyle="none", c='k', mew=2)

for plotting the RF threshold, but this should be

plt.plot(fpr_rf[close_default_rf], tpr_rf[close_default_rf], '^', markersize=10,
         label="threshold 0.5 RF", fillstyle="none", c='k', mew=2)

instead, right?

[joaquinvanschoren]

Yes indeed!
Thanks for noticing it. I fixed it now.