TensPLS_fit: Tensor envelope partial least squares (PLS) regression

In kusakehan/TEReg: An R package for tensor envelope regression with tensor response and tensor predictor

Description

This function estimates the factor matrix W_k, k=1,\cdots,m in tensor PLS algorithm, see Zhang, X., & Li, L. (2017).

Usage

TensPLS_fit(Xn, Yn, SigX, u)

Arguments

Xn	A predictor tensor of dimension p_1\times \cdots \times p_m.
Yn	The response vector of dimension r.
SigX	A matrix lists \boldsymbol{Σ}_k, k=1,\cdots, m, which determins the estimation of covariance matrix \boldsymbol{Σ}=\boldsymbol{Σ}_m \otimes \cdots \otimes \boldsymbol{Σ}_1.
u	The dimension of envelope subspace, u=(u_1,\cdots,u_m).

Value

Gamma	The estimation of factor matrix W_k, k=1,\cdots,m.
PGamma	The projection matrix W_k(W_k'\boldsymbol{Σ}_k W_k)^{(-1)}W_k'\boldsymbol{Σ}_k, k=1,\cdots,m.

References

Zhang, X., & Li, L. (2017). Tensor Envelope Partial Least-Squares Regression. Technometrics, 59(4), 426-436.

kusakehan/TEReg documentation built on May 30, 2019, 7:17 a.m.