trambakbanerjee/casp: Coordinate-wise Adaptive Shrinkage Prediction

Provides the CASP rule for shrinkage prediction in high-dimensional, non-exchangeable hierarchical Gaussian models with an unknown location as well as an unknown spiked covariance structure. CASP utilizes the phase transition phenomenon of the sample eigenvalues and eigenvectors seen in spiked covariance models and improves upon naive factor model based methodology by using bias-corrected efficient estimates of quadratic forms involving the covariance matrix that appear in the Bayes predictive rules.

install.packages("remotes")
remotes::install_github("trambakbanerjee/casp")