LogOddsFusionResult#
- class scikitplot.corpus.LogOddsFusionResult(items, evidence_correlation, shared_counts)[source]#
A fused ranking and the diagnostics needed to trust it.
- Parameters:
- itemstuple of LogOddsFusedItem
Descending log-odds, ties by key, at most
top_k.- evidence_correlationdict
For every pair of legs
(a, b)(positions,a < b): marginal Pearson correlation of their evidence over the items both listed, orNonewhen fewer than three items are shared or either side is constant. Not a test of conditional independence (see the module notes); near 1 means near-duplicate legs.- shared_countsdict
For the same pairs, how many items both legs listed.
- Parameters:
- items: tuple[LogOddsFusedItem, ...][source]#