Judge how well a machine learning algorithm fits

A classification model is judged on a confusion matrix of true and false positives and negatives. Precision, TP ÷ (TP + FP), matters when false positives are costly. Recall, or sensitivity, TP ÷ (TP + FN), matters when missing a positive is costly. Accuracy, (TP + TN) ÷ all observations, can mislead when classes are unbalanced. The F1 score, 2 × precision × recall ÷ (precision + recall), balances the two and suits unbalanced data. A receiver operating characteristic curve plots the true positive rate against the false positive rate across thresholds; a larger area under the curve generally means a better model. For continuous targets, root mean squared error is the usual measure.