Accuracy
Share of correct predictions.
← Machine Learning learning path0%
08
Supervised learning
Model Evaluation
Measuring classifier quality honestly and comparing models fairly.
Not started yet
Start Quiz mastery0%
What you’ll take away
- Every classification metric comes from the confusion matrix.
- Accuracy misleads on imbalanced data; check the majority-class baseline.
- Precision counts false alarms, recall counts misses; F1 balances both.
- ROC-AUC summarises all thresholds; cross-validation gives stable estimates.
Major concepts (syllabus)
Accuracy, precision, recall, F1-score, ROC/AUC, confusion matrices, cross-validation
Lessons
- Confusion matricesTP, FP, TN and FN in one table.8 min
- AccuracySimple, and misleading on imbalanced data.6 min
- Precision and recallFalse alarms versus missed positives.9 min
- F1-scoreThe harmonic mean of precision and recall.7 min
- ROC curves and AUCPerformance across every threshold.10 min
- Cross-validationk-fold estimates that use all the data for both training and validation.9 min
Key formulas
Precision
Of predicted positives, how many are right.
Recall
Of actual positives, how many were found.
F1-score
Harmonic mean of precision P and recall R.
True / false positive rate
The two axes of the ROC curve.
Key terms
- Accuracy
- Fraction of predictions that are correct. Misleading when classes are imbalanced.
- AUC
- Area under the ROC curve: probability a random positive is ranked above a random negative. 0.5 is random.
- Confusion matrix
- Table of predicted vs. actual classes: TP, FP, TN, FN for binary problems.
- Cross-validation
- Rotating which fold is held out for validation and averaging the scores.
- F1-score
- Harmonic mean of precision and recall.
- Precision
- TP / (TP + FP): share of predicted positives that are correct.
- Recall
- TP / (TP + FN): share of actual positives found.
- ROC curve
- True positive rate vs. false positive rate across all thresholds.
Practice this topic6 questions · earn XP and build your streak