precision | (must be at least ) | |

recall | (can be at most ) |

True proportion of 1s in the data

Return matrix

predicted | |||

0 | 1 | ||

true | 0 | ||

1 |

Expected return of one prediction is **.**

Reference return

Probability of net loss on 1000 predictions: **.**

Copyright Antti Ukkonen 2019, released under the MIT license.