s3faInit: Generate initial parameters for EM/S3FA

In aciobanusebi/s2fa: Factor Analysis (FA), Simple Supervised FA (S2FA), Simple Semi-Supervised FA (S3FA), Missing S3FA (MS3FA)

Description

Generate initial parameters for EM/S3FA

Usage

s3faInit(X_t_supervised, Z_t_supervised, X_t_unsupervised, type = "fa",
  checkArgs = TRUE)

Arguments

X_t_supervised	train input data (which has output) as design matrix (must be matrix, not data.frame), i.e. row = instance, column = feature/attribute
Z_t_supervised	train output data (for X_t_supervised) as design matrix (must be matrix, not data.frame), i.e. row = instance, column = feature/attribute
X_t_unsupervised	train input data (which has no output) as design matrix (must be matrix, not data.frame), i.e. row = instance, column = feature/attribute
type	"unconstrained", "fa", or "ppca"; refers to psi
checkArgs	whether to check the arguments are valid; it takes more time to execute

Value

initial parameters for EM/S3FA, i.e. a list containing nDimX, nDimZ, type, mu_z_t, Sigma_z_t, mu_t, lambda_t, psi_t. "_t" comes from "transpose"

Examples

params <- s3faInit(X_t_supervised = house[1:10,2:3,drop=FALSE],
                   Z_t_supervised = house[1:10,1,drop=FALSE],
                   X_t_unsupervised = house[11:20,2:3,drop=FALSE],
                   type = "fa",
                   checkArgs = TRUE)
params

aciobanusebi/s2fa documentation built on Aug. 7, 2021, 6:38 a.m.