| MMboot_loccov {FRB} | R Documentation |
Calculates bootstrapped MM-estimates of multivariate location and scatter using the Fast and Robust Bootstrap method.
MMboot_loccov(Y, R, ests = MMest_loccov(Y))
Y |
matrix or data frame |
R |
number of bootstrap samples |
ests |
original MM-estimates as returned by MMest_loccov() |
This function is called by FRBpcaMM and FRBhotellingMM, it is typically not to be used on its own.
It requires the result of MMest_loccov applied on Y, supplied through the argument ests.
If ests is not provided, MMest_loccov will be called with default arguments.
The fast and robust bootstrap was first developed by Salibian-Barrera and Zamar (2002) for univariate regression MM-estimators.
The value centered gives a matrix with R columns and 2*(p+p*p) rows (p is the number of variables in Y),
containing the recalculated estimates of the MM-location, MM-shape, S-covariance and S-location.
Each column represents a different bootstrap sample.
The first p rows are the MM-location estimates, the next p*p rows are the MM-shape estimates (vectorized). Then the next
p*p rows are the S-covariance estimates (vectorized) and the final p rows are the S-location estimates.
The estimates are centered by the original estimates, which are also returned through MMest in vectorized form.
A list containing:
centered |
recalculated MM- and S-estimates of location and scatter (centered by original estimates), see Details |
MMest |
original MM- and S-estimates of location and scatter, see Details |
Gert Willems and Ella Roelant
FRBpcaMM, FRBhotellingMM, MMest_loccov, Sboot_loccov
Y <- matrix(rnorm(50*5), ncol=5) MMests <- MMest_loccov(Y) bootresult <- MMboot_loccov(Y, R = 1000, ests = MMests)