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.

The mean Poisson deviance — sklearn.metrics.mean_poisson_deviance.

public static double Score(ReadOnlySpan<double> yTrue, ReadOnlySpan<double> yPred, ReadOnlySpan<double> sampleWeight = default)

Parameters — yTrue is the true counts, which must be non-negative, and yPred the predicted rates, which must be strictly positive and the same length. sampleWeight is one weight per sample, or empty — the default — to weight every sample by 1.

Returns — double, 0 for a perfect prediction and larger the worse it is. Unbounded above.

Exceptions — ArgumentException when the lengths disagree, the input is empty or holds a non-finite value, a truth is negative, or a prediction is not strictly positive. The message is scikit-learn's, and names power=1: this is TweedieDeviance at that power and shares its refusal.

Example — four counts against a model's predicted rates.

using Lodestar.Metrics;

double[] counts = [1.0, 2.0, 3.0, 4.0];
double[] rates = [1.5, 2.5, 2.0, 4.5];

double deviance = PoissonDeviance.Score(counts, rates);  // => 0.1967…

A zero count is fine, and contributes only the ŷ − y part of the term:

using Lodestar.Metrics;

double[] counts = [0.0, 2.0, 3.0];
double[] rates = [1.0, 2.0, 3.0];

double withZero = PoissonDeviance.Score(counts, rates);  // => 0.6666…

Remarks — identical to TweedieDeviance.Score at power: 1.0, asserted across the whole frozen corpus rather than on one pair.

Applies to — net10.0, netstandard2.0.

See also — GammaDeviance.Score, TweedieDeviance.Score, D2Tweedie.Score, the Python equivalence table.