Table 2.

Side-by-Side Comparison of Hypothesis Testing and Binary Classification

	Hypothesis Testing	Binary Classification
Symmetry between binary answers	Asymmetric (default is 0)	Symmetric or asymmetric
No. of instances to make one decision given a decision rule	$\ge 1$ (the larger the better)	$=1$
Available binary answers	No	Yes (training data)
Evaluation criteria	Power (given a significance level)	Prediction accuracy
With the largest possible no. of instances	Power = 1	Prediction accuracy not necessarily perfect