roiMLE: Compute the MLE for a selected region of interest
In ammeir2/selectiveROI:

Description Usage Arguments

View source: R/newMLEfunction.R

Computes the conditional mle for a region selected based on the selection rule y[selected] > threshold or y[selected] < -threshold, and the coordinates which were not selected must violate the selection rule.

roiMLE(y, cov, threshold, compute = c("mle", "lower-CI", "upper-CI"),
  ci_alpha = 0.025, coordinates = NULL, selected = NULL,
  mean_weights = NULL, projected = NULL, regularization_param = NULL,
  regularization_slack = 1, init = NULL, progress = FALSE,
  sampling_control = roi_sampling_control(),
  mle_control = roi_mle_control())

`y`	the observed noraml coordinates
`cov`	the covariance of `y`
`threshold`	the threshold used in the selection rule. Must be either a scalar or a numeric vector of size `length(y)`
`coordinates`	an optional matrix of the coordinates of the observed vector. This is only relevant if `y` corresponds to a spatial or temporal observation
`selected`	an optional boolean vector, with `TRUE` coordinates corresponding to coordinates of `y` that were selected.
`mean_weights`	the weights to use for the contrast to be computed, if not specified then equal weights will be given to all selected coordinates. `mean_weights` do not have to sum to one. If specified, then the length of the vector must either equal `sum(selected)` (the number of selected coordinates), or `length(y)`. Either way, all coordinates which were not selected must be given a weight of zero.
`projected`	an optional fixed value that `mean(mu)` must equal. Can be used to construct profile likelihood post-selection confidence intervals.
`regularization_param`	an optional penalty value for the tykohonov regularizer. This is an (inferior) altenative to specifying a `regularization_slack`.
`regularization_slack`	the estimation routine uses first differences Tykohonov regularization to estimate the mean of the selected region. `regularization_slack` speficies the allowed deviation from the observed first order differences. The description for details
`init`	initial value for the mean estimate
`progress`	whether to display a bar describing the progress of the gradient algorithm.
`sampling_control`	a list with control parameters for sampling from the estimated distribution.
`mle_contorl`	a list of parameters to be used when computed the conditional MLE.