pystra.active_learning.estimation.MonteCarloEstimator#

Use it: User guide · Worked example · Theory

class MonteCarloEstimator(*, n_samples=100000)[source]#

Bases: ReliabilityEstimator

Independent normal Monte Carlo with an exact 95 % binomial interval.

n_samples defaults to 100000 and must be at least 2. No/all failures yield infinite CoV, so endpoint samples cannot claim adequate precision. All diagnostics are conditional on the supplied surrogate.

Methods

estimate

Classify fresh row-wise normal points using the supplied predictor.

estimate(predict, *, dimension, rng)[source]#

Classify fresh row-wise normal points using the supplied predictor.

Return type:: ReliabilityEstimate