pystra.active_learning.subset.SubsetLevel

Use it: User guide · Worked example · Theory

class SubsetLevel(threshold, conditional_probability, n_samples, n_chains, variance_factor, acceptance_rate)

Bases: object

One level’s threshold, conditional probability and chain diagnostics.

The final probability concerns mean <= 0, while exploratory intermediate levels condition on mean - 2 std. variance_factor compares a chain cluster variance estimate with IID Bernoulli variance, floored at one. It does not account for ancestry shared by chains at different levels. acceptance_rate concerns proposals generating this level (one at the initial IID level). Seeds are included in each conditional chain.

Methods