Classification

Classification predicts a discrete category from observed features. Biological examples include disease status, cell type or treatment-response group.

Probability and decision

A binary classifier may estimate

A threshold then converts this probability into a class prediction. Changing the threshold changes the balance between false positives and false negatives.

Evaluation

Accuracy alone can be misleading when classes are imbalanced. Sensitivity, specificity, precision, recall and ROC-based measures provide different views of performance.

Biological consequences

The appropriate metric and threshold depend on the consequences of different errors, not only on mathematical convenience.