Expand description
Generalized covariance measure (GCM) conditional-independence test (continuous data), after Shah & Peters (2020).
X and Y are linearly regressed on [1, Z]; with residuals r_x, r_y and
u = r_x ∘ r_y, the statistic is T = Σu / (√n · std(u)) (population
standard deviation), asymptotically N(0, 1) under H₀. The p-value is
2 · P(N > |T|) and the effect size is |cor(r_x, r_y)|.
Structs§
- Gcm
- Generalized covariance measure test (linear residualization).