Statistical Inference for Signed-Bottleneck NMF-AE Signed Bottleneck

Post-estimation inference for the signed bottleneck \(\Theta = C_{+} - C_{-}\) in the Signed-Bottleneck NMF-AE model \(Y_1 \approx X_1 \Theta X_2 Y_2\), conditional on \((\hat X_1, \hat X_2)\). Uses sandwich covariance and wild bootstrap without the non-negativity projection that nmfae.inference applies (because \(\Theta\) is unconstrained in sign here).

Usage

nmfae.signed.inference(object, Y1, Y2 = Y1, wild.bootstrap = TRUE, ...)

Arguments

object

A fitted "nmfae.signed" object.

Y1

Output matrix used during fitting.

Y2

Input matrix used during fitting. Default Y1.

wild.bootstrap

Logical. Default TRUE.

...

Additional arguments:

wild.B: Bootstrap replicates. Default 500.
wild.seed: RNG seed. Default 123.
wild.level: CI confidence level. Default 0.95.
sandwich: Use sandwich covariance. Default TRUE.
C.p.side: P-value type: "two.sided" (default for Signed-Bottleneck NMF-AE) or "one.sided".
cov.ridge: Ridge stabilization. Default 1e-8.
print.trace: Logical. Default FALSE.

Value

An object of class c("nmfae.signed.inference", "nmfae.inference", "nmfae.signed", "nmfae", "nmf") with added fields:

sigma2.used: Estimated \(\sigma^2\).
C.se, C.se.boot: Sandwich / bootstrap SEs for \(\Theta\) (Q x R).
C.ci.lower, C.ci.upper: Bootstrap CIs.
coefficients: Data frame with Estimate, SE, BSE, z, p-value, CI.
C.p.side: P-value side used.

Lifecycle

This function is experimental. The interface may change in future versions.

References

Ding, C. H. Q., Li, T., & Jordan, M. I. (2010). Convex and semi-nonnegative matrix factorizations. IEEE Transactions on Pattern Analysis and Machine Intelligence, 32(1), 45–55.