perform_shrink: performs shrinkage estimation and use Extended BIC (EBIC) to...
In zhenkewu/slamR: Structured Latent Attribute Models in R

performs shrinkage estimation and use Extended BIC (EBIC) to select the best tuning parameter in the penalized likelihood from an entire solution path

perform_shrink(
  X,
  Q_est,
  Z_candi,
  lambda_vec,
  c_ini,
  g_ini,
  is_sub = 0,
  model = "DINA"
)

`X`	N by J data matrix
`Q_est`	J by K Q after screening by `adg_em` (one-level version)
`Z_candi`	K-column candidate latent attributes (candidates; `adg_em` (one-level version) )
`lambda_vec`	a decreasing sequence of negative penalty parameters
`c_ini`	J initial values of 1-slipping parameters
`g_ini`	J initial values of guessing parameters
`is_sub`	use subsample (1) or not (0; default) - NB: not implemented.
`model`	"DINA" (default) or "DINO"

A_final final estimated set of latent attribute patterns that made the cut
EBIC_vec vector of EBIC values
size_vec for each tunning parameter, the size of latent attribute patterns that made the cut.

zhenkewu/slamR documentation built on March 8, 2020, 1:31 a.m.