peopleanalyst

Platform · Live · new here

True variance decomposition — driver importance under predictor intercorrelation

HO-384: Johnson (2000) relative weights (Jacobi eigendecomposition, k<=50) + exact general dominance / Shapley of subset R-squared (Gaussian elimination, k<=12) over a caller-supplied predictor intercorrelation matrix + per-driver validities. Replaces the independence-assumption r-squared apportionment principia sends (its intercorrelation_matrix.available=false by design) with the TRUE decomposition when a matrix exists. 3-state honesty: computed-true / independence-assumed (labeled, never silent) / refused (9 named codes incl. not-positive-definite). Integrated into portfolio-voi: optional matrix swaps naive apportionment for epsilon-weight shares before ranking; response carries decompositionMethod. Two independent computations (eigen vs Shapley) agree to 1e-9 in tests. Upgrade trigger: PRN-189 option b (principia supplies the pairwise matrix).

Cluster
Platform
Type
algo
Status
Live
Used by
people-analytics-toolbox

Maturity evidence

Functioning
18 node:test tests green: relative weights sum to hand-computed R-squared=0.2981099656 (Cramer's rule), eigen path vs Shapley path agree to 1e-9 (two independent computations), orthogonal matrix reduces to naive apportionment, correlated fixture diverges in the asserted direction, 9 refusal codes exercised, portfolio-voi rank-flip fixture proven. Contract 1.6.0; REST + MCP registered.
Valuable
No consumer supplies a matrix yet; value lands when a consumer computes intercorrelations from its own data or PRN-189 option b (principia supplies the matrix) ships.
Understood
Method, caps, and the independence-assumption boundary documented in the module header, MCP description, and this entry.
Integrated
Wired INTO portfolio-voi (the existing HO-370 consumer path): matrix present changes the ranking and the response labels its method — both doors exposed, discovery drift fixed (portfolio-voi was missing from the MCP discovery list).

How to use it

REST
/api/spokes/forecasting/variance-decomposition
MCP
forecasting.variance-decomposition
Data
src/spokes/forecasting/core/variance-decomposition.ts

Other Platform capabilities

← Every platform capability

Capability detail sourced from the People Analytics Toolbox capability feed (source of truth), snapshot retrieved 2026-07-23.