Skip to main content

Module indexed_response

Module indexed_response 

Source
Expand description

Structural geometry and likelihood measures for separable indexed responses.

A response cell can be structurally absent (missing, not at risk, or outside a declared output domain) independently of the numerical weight attached to an observed cell. Conflating those concepts in a single zero-valued weight loses the response geometry and makes it impossible to validate event/risk sets or preserve them in a fitted model. This module keeps them distinct.

Structs§

IndexedCellSet
A sorted, duplicate-free set of (row, output) cells in CSR form.
IndexedResponseError
Invalid indexed-response geometry or likelihood measure.
OwnedSeparableCellMeasure
Lifetime-free structural geometry and likelihood measure.
SeparableCellMeasure
Structural activity plus numerical likelihood weights for a separable response. An inactive cell has no likelihood contribution; an active cell with weight zero remains an observed member of the response geometry.

Enums§

LikelihoodWeights
Numerical likelihood weighting, independent of structural presence.
OwnedCellValues
Owned values over a declared (row, output) grid.
OwnedLikelihoodWeights
Owned numerical likelihood measure, independent of structural presence.
OwnedStructuralCells
Owned structural presence for a separable response grid.
StructuralCells
Structural presence of cells in a separable response grid.