Lodestar.Metrics 0.3.0. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

ZeroOneLoss

The fraction of samples predicted wrongly — one minus Accuracy on single-label input, and something else entirely on a label matrix.

There, a row counts as wrong if any of its labels differs. One wrong label out of three costs a whole sample, where HammingLoss charges a third of one. Measured on two samples over three labels with two labels wrong, one in each row: 1 here and 0.3333… there.

normalize follows Accuracy.Score's: pass false and the answer is the weight of the wrong samples rather than their share — a count when every weight is 1.

Members

Member	What it does
ZeroOneLoss.Score	The share of samples that disagree, from labels or from a matrix.