findThresholdSelect: Find Optimal Threshold for Amplified, Initially Marginal,...
In dajmcdon/aimer: Amplified, Initially Marginal, Eigenvector Regression

Description Usage Arguments Value

find the optimal number of covariates, number of components, and number of covariates in selection process for AIMER method using kfold cross-validation.

findThresholdSelect(X, y, ncomps, nCovs = NULL,
  nCovs.min = ifelse(is.null(nCovs), max(ncomps) + 2, min(nCovs)),
  nCovs.max = ifelse(is.null(nCovs), nrow(X), max(nCovs)),
  nthresh = ifelse(is.null(nCovs), 25, length(nCovs)), nCovs.select = NULL,
  nCovs.min.select = ifelse(is.null(nCovs.select), max(ncomps) + 2,
  min(nCovs.select)), nCovs.max.select = ifelse(is.null(nCovs.select),
  nrow(X), max(nCovs.select)), nthresh.select = ifelse(is.null(nCovs.select),
  25, length(nCovs.select)), kfold = 10)

`X`	required, design matrix with dimension (n,p).
`y`	required, response vector with dimension n.
`ncomps`	required, number of components, can be an integer or a vector of integers.
`nCovs`	optional, a vector of possible numbers of covariates.
`nCovs.min`	optional, the smallest number of covariates, default as max(`ncomps`)+2.
`nCovs.max`	optional, the largest number of covariates, default as number of rows of `x`.
`nthresh`	optional, how many `nCovs` to be tested, default as 25.
`nCovs.select`	optional, a vector of possible numbers of covariates in selection process.
`nCovs.min.select`	optional, the smallest number of covariates in selection process, default as max(`ncomps`)+2.
`nCovs.max.select`	optional, the largest number of covariates in selection process, default as number of rows of `x`.
`nthresh.select`	optional, how many `nCovs.select` to be tested in selection process, default as 25.
`kfold`	required, the number of k in kfold cross-validation, default as 10.

an object of class 'supervisedPCACV', a list with the following components

`nCov.select.best`	the best number of covariates in selection process which gives smallest mse in cross-validation
`nCov.best`	the best number of covariates which gives smallest mse in cross-validation
`ncomp.best`	the best number of components which gives smallest mse in cross-validation
`nCovs.select`	all the tested numbers of covariates in selection process
`ncomps`	all the tested ncomps
`nCovs`	all the tested numbers of covariates
`mse`	average mse in cross-validation over k folds