Computes the leverage of each observation in the PC score / IC mixing matrix (U/M)
for projection scrubbing. Optionally can threshold the leverage values to
flag potential outliers.
leverage(Comps, are_orthogonal = FALSE, median_cutoff = NULL)
Arguments
| Comps |
The \(n\) by \(Q\) PC score matrix/IC mixing matrix. |
|---|
| are_orthogonal |
Can the columns of Comps be assumed to be orthogonal
and have a 2-norm of 1? Speeds up the computation. |
|---|
| median_cutoff |
The outlier cutoff, in multiples of the median leverage.
Default: NULL (do not compute outliers). |
|---|
Value
A list with entries "meas" (the leverage values),
"cut" (the leverage cutoff value) and
"flag" (logical vector indicating the outliers). If
!is.null(median_cutoff), "cut" and "flag" are omitted.