Metrics from a confusion matrix
Fill in a confusion matrix and watch accuracy lie on imbalanced data. Then read four sets of training curves the way an examiner expects you to.
Predict first
Commit to a guess before you touch anything. Being wrong here is how it sticks.
TP = 40, FP = 10, FN = 20. What's the F1 score? (2 dp)
Experiment
Move the controls and watch the outputs. Goals tick off automatically when you reach them.
Metrics from a confusion matrix
Read the curves
Training loss keeps falling while validation rises after ~30% of training: overfitting (high variance). Augment, regularise, stop early, or get more data.
Explain it
Put what you saw into words. This is what the exam will ask you to do.
How do you diagnose overfitting vs underfitting from loss curves, and what do you do about each?
Code it
Write real code against real data. Python and SQL run right here in your browser and hidden tests check your work; the AI reviews it like an examiner.
Metrics from predictions
Write metrics(y_true, y_pred) (NumPy arrays of 0/1) returning a dict with tp, fp, fn, tn,
accuracy, precision, recall, f1 — use 0.0 when a denominator is 0.
Challenge
Fixed variations, a new problem every day, and unlimited AI challenges built from your current settings.
Harmonic Mean F1 Score
On a medical test set: TP = 40, FP = 10, FN = 20, TN = 930. What is the F1 score (as a decimal between 0 and 1, rounded to 2 decimals)?