Description Usage Arguments Details Value Author(s) See Also Examples
Function to compute optimally robust estimates for L2-differentiable parametric families via k-step construction.
1 2 3 4 5 | robest(x, L2Fam, fsCor = 1, risk = asMSE(), steps = 1L,
verbose = NULL, OptOrIter = "iterate", nbCtrl = gennbCtrl(),
startCtrl = genstartCtrl(), startICCtrl = genstartICCtrl(),
kStepCtrl = genkStepCtrl(), na.rm = TRUE, ..., debug = FALSE,
withTimings = FALSE, diagnostic = FALSE)
|
x |
sample |
L2Fam |
object of class |
fsCor |
positive real: factor used to correct the neighborhood radius; see details. |
risk |
object of class |
steps |
positive integer: number of steps used for k-steps construction |
verbose |
logical: if |
OptOrIter |
character; which method to be used for determining Lagrange
multipliers |
nbCtrl |
a list specifying input concerning the used neighborhood;
to be generated by a respective call to |
startCtrl |
a list specifying input concerning the used starting estimator;
to be generated by a respective call to |
startICCtrl |
a list specifying input concerning the call to
|
kStepCtrl |
a list specifying input concerning the used variant of
a kstepEstimator;
to be generated by a respective call to |
na.rm |
logical: if |
... |
further arguments |
debug |
logical: if |
withTimings |
logical: if |
diagnostic |
logical; if |
A new, more structured interface to the former function roptest
.
For details, see this function.
In some respects this functions allows for more granular arguments,
in the sense that the different steps (a) computation of the inital estimator,
resp. (a') in case initial.est
is missing computation of the initial
MDE, (b) computation of the optimal IC and (c) computation of the k-step
estimator each can have individial arguments E.arglist
to be
passed on to calls to expectation operator E
within each step.
These different arguments are passed through the input generating functions
genstartCtrl
,
genstartICCtrl
, and
kStepCtrl
Diagnostics on the involved integrations are available if argument
diagnostic
is TRUE
. Then there are attributes diagnostic
and kStepDiagnostic
attached to the return value, which may be inspected
and assessed through showDiagnostic
and
getDiagnostic
.
Object of class "kStepEstimate"
. In addition, it has
an attribute "timings"
where computation time is stored.
Matthias Kohl Matthias.Kohl@stamats.de,
Peter Ruckdeschel peter.ruckdeschel@uni-oldenburg.de
roblox
,
L2ParamFamily-class
UncondNeighborhood-class
,
RiskType-class
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 | ## Don't test to reduce check time on CRAN
#############################
## 1. Binomial data
#############################
## generate a sample of contaminated data
set.seed(123)
ind <- rbinom(100, size=1, prob=0.05)
x <- rbinom(100, size=25, prob=(1-ind)*0.25 + ind*0.9)
## Family
BF <- BinomFamily(size = 25)
## ML-estimate
MLest <- MLEstimator(x, BF)
estimate(MLest)
confint(MLest)
## compute optimally robust estimator (known contamination)
nb <- gennbCtrl(eps=0.05)
robest1 <- robest(x, BF, nbCtrl = nb, steps = 3)
estimate(robest1)
confint(robest1, method = symmetricBias())
## neglecting bias
confint(robest1)
plot(pIC(robest1))
tmp <- qqplot(x, robest1, cex.pch=1.5, exp.cex2.pch = -.25,
exp.fadcol.pch = .55, jit.fac=.9)
## compute optimally robust estimator (unknown contamination)
nb2 <- gennbCtrl(eps.lower = 0, eps.upper = 0.2)
robest2 <- robest(x, BF, nbCtrl = nb2, steps = 3)
estimate(robest2)
confint(robest2, method = symmetricBias())
plot(pIC(robest2))
## total variation neighborhoods (known deviation)
nb3 <- gennbCtrl(eps = 0.025, neighbor = TotalVarNeighborhood())
robest3 <- robest(x, BF, nbCtrl = nb3, steps = 3)
estimate(robest3)
confint(robest3, method = symmetricBias())
plot(pIC(robest3))
## total variation neighborhoods (unknown deviation)
nb4 <- gennbCtrl(eps.lower = 0, eps.upper = 0.1,
neighbor = TotalVarNeighborhood())
robest3 <- robest(x, BF, nbCtrl = nb4, steps = 3)
robest4 <- robest(x, BinomFamily(size = 25), nbCtrl = nb4, steps = 3)
estimate(robest4)
confint(robest4, method = symmetricBias())
plot(pIC(robest4))
#############################
## 2. Poisson data
#############################
## Example: Rutherford-Geiger (1910); cf. Feller~(1968), Section VI.7 (a)
x <- c(rep(0, 57), rep(1, 203), rep(2, 383), rep(3, 525), rep(4, 532),
rep(5, 408), rep(6, 273), rep(7, 139), rep(8, 45), rep(9, 27),
rep(10, 10), rep(11, 4), rep(12, 0), rep(13, 1), rep(14, 1))
## Family
PF <- PoisFamily()
## ML-estimate
MLest <- MLEstimator(x, PF)
estimate(MLest)
confint(MLest)
## compute optimally robust estimator (unknown contamination)
nb1 <- gennbCtrl(eps.upper = 0.1)
robest <- robest(x, PF, nbCtrl = nb1, steps = 3)
estimate(robest)
confint(robest, symmetricBias())
plot(pIC(robest))
tmp <- qqplot(x, robest, cex.pch=1.5, exp.cex2.pch = -.25,
exp.fadcol.pch = .55, jit.fac=.9)
## total variation neighborhoods (unknown deviation)
nb2 <- gennbCtrl(eps.upper = 0.05, neighbor = TotalVarNeighborhood())
robest1 <- robest(x, PF, nbCtrl = nb2, steps = 3)
estimate(robest1)
confint(robest1, symmetricBias())
plot(pIC(robest1))
#############################
## 3. Normal (Gaussian) location and scale
#############################
## 24 determinations of copper in wholemeal flour
library(MASS)
data(chem)
plot(chem, main = "copper in wholemeal flour", pch = 20)
## Family
NF <- NormLocationScaleFamily()
## ML-estimate
MLest <- MLEstimator(chem, NF)
estimate(MLest)
confint(MLest)
## Don't run to reduce check time on CRAN
## Not run:
## compute optimally robust estimator (known contamination)
## takes some time -> you can use package RobLox for normal
## location and scale which is optimized for speed
nb1 <- gennbCtrl(eps = 0.05)
robEst <- robest(chem, NF, nbCtrl = nb1, steps = 3)
estimate.call(robEst)
attr(robEst,"timings")
estimate(robest)
confint(robest, symmetricBias())
plot(pIC(robest))
## plot of relative and absolute information; cf. Kohl (2005)
infoPlot(pIC(robest))
tmp <- qqplot(chem, robest, cex.pch=1.5, exp.cex2.pch = -.25,
exp.fadcol.pch = .55, withLab = TRUE, which.Order=1:4,
exp.cex2.lbl = .12,exp.fadcol.lbl = .45,
nosym.pCI = TRUE, adj.lbl=c(1.7,.2),
exact.pCI = FALSE, log ="xy")
## finite-sample correction
if(require(RobLox)){
n <- length(chem)
r <- 0.05*sqrt(n)
r.fi <- finiteSampleCorrection(n = n, r = r)
fsCor0 <- r.fi/r
nb1 <- gennbCtrl(eps = 0.05)
robest <- robest(chem, NF, nbCtrl = nb1, fsCor = fsCor0, steps = 3)
estimate(robest)
}
## compute optimally robust estimator (unknown contamination)
## takes some time -> use package RobLox!
nb2 <- gennbCtrl(eps.lower = 0.05, eps.upper = 0.1)
robest1 <- robest(chem, NF, nbCtrl = nb2, steps = 3)
estimate(robest1)
confint(robest1, symmetricBias())
plot(pIC(robest1))
## plot of relative and absolute information; cf. Kohl (2005)
infoPlot(pIC(robest1))
## End(Not run)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.