Description Usage Arguments Details Value Warning Note Author(s) See Also Examples
This function adds confidence calculations to
various entities in abrem
objects.
1  abrem.conf(x,which="all",...)

x 
Object of class 
which 
Calculate which fit in the 
... 
Options for calculating confidence, and for plotting the results. 
This function adds confidence calculations to various entities in
abrem
objects and adds them to the object alongside any preexisting
confidence calculations.
Additional options for calculating Blife confidence are passed with:
cl
Confidence level: A single number from the interval [0,[1
specifying the confidence level for various confidence calculations.
Defaults to 0.9
.
conf.blives.sides
Either "lower"
, "upper"
or "double"
,
specifying the type of bound(s) to be calculated.
Defaults to c("double")
, the other options are currently
not implemented.
unrel.n
An integer controlling the amount of unreliability levels for which Blife confidence bounds are calculated and ultimately plotted.
Higher numbers will result in smoother confidence bounds. In any case, confidence intervals will be calculated for:
the Blives at unreliability levels specified with option unrel
the Blife at 50 [%]
unreliability
the Blife at the calculcate characteristic life or logmean (depending on the fitted distribution)
Note: When plotting fits and confidence bounds that are adjusted with
a threshold (see option "threshold"
), it is often the case that
the bounds appear to be cut of on the left. This can be countered by
dramatically increasing unrel.n
, resulting in confidence
bounds that extend to the edge of the plotting area.
Defaults to 25
.
conf.what
A vector of class "character"
describing for which entities
that confidence should be calculated.
Defaults to c("blives")
, the only type currently supported.
unrel
An unordered numeric vector with unreliability levels for which Blife confidence will be calculated.
Defaults to c(0.1,0.05,0.01)
.
method.conf.blives
A vector of class "character"
describing the technique to be
used for calculating confidence for Blives. Possible values are
"bbb"
(Beta Binomial confidence bounds),
"lrb"
(Likelihood Ratio confidence bounds) and
"mcpivotals"
or "mcpivotal"
(Monte Carlo Pivotal
confidence bounds).
Monte Carlo Pivotal confidence bounds use a large number of
simulations to calculate the confidence bounds. See option
"S"
for more info.
Defaults to c("mcpivotals")
.
S
An integer describing the number of Monte Carlo simulations on which the Monte Carlo pivotal confidence bounds and calculation of the "prr" goodnessoffit indicator are based.
High values are needed for good confidence bounds at the lower end of the fitted model, especially for data with heavy censoring.
Note that S >= 100
and that S
must be divisible by 10.
Defaults to 10000
.
in.legend
Logical value controlling the inclusion of confidence calculation results in the legend.
If in.legend=FALSE
is passed ,
the resulting confidence calculations will be omitted from the legend.
Defaults to TRUE
.
Additionally, one can pass any options available from options.abrem
,
such as col
or is.plot.legend
. The graphical options
will be used when plotting the (life)time observations using plot.abrem
.
The function returns its argument x
, extended with the confidence
calculations and any optional graphical and calculation arguments
as passed to the function.
Currently, the Monte Carlo pivotal confidence bounds are only identical to superSMITH's MC pivotal bounds for complete, uncensored data. For heavily censored datasets with few failures, the bounds appear more optimistic than superSMITH's bounds. Research on this issue is ongoing.
Currently, only which = "all"
is supported, meaning that a
call to abrem.conf
attempts calculation of confidence for all
fits in the abrem
object.
Currently, only conf.what = "blives"
and
conf.blives.sides = "double"
are supported.
Jurgen Symynck [email protected]
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  ## full dataset ##
da1 < Abrem(runif(10,100,1e4),label="Complete data")
da1 < abrem.fit(da1)
da1 < abrem.conf(da1,method.conf.blives="mcpivotals",col="red")
da1 < abrem.conf(da1,method.conf.blives="bbb",col="orange")
da1 < abrem.conf(da1,method.conf.blives="lrb",col="yellow3")
print(da1$fit[[1]]$conf$blives[[1]])
plot(da1,main="Comparison between MC Pivotal bounds and BB Bounds")
## censored dataset: generates a warning for MC Pivotal confidence bounds ##
da2 < runif(8,100,1e4)
da2 < Abrem(fail=da2,susp=rep(max(da2),2),label="Type II censored data")
# generate a 'type 2' censored dataset
da2 < abrem.fit(da2)
da2 < abrem.conf(da2,method.conf.blives="mcpivotals",col="blue1")
da2 < abrem.conf(da2,method.conf.blives="bbb",col="steelblue")
da2 < abrem.conf(da2,method.conf.blives="lrb",col="cyan3")
plot(da2,main="Comparison between different bound types.")
## show variability in Monte Carlo Pivotal bounds with low S ##
da3 < Abrem(rweibull(5,3,1000))
da3 < abrem.fit(da3)
for(i in 1:20) da3 < abrem.conf(da3,S=1000,lwd=1,col="red")
# just keep adding bounds to the abrem object...
plot(da3,is.plot.legend=FALSE,
main="Variability in MC Pivotal Conf. Bounds for S=1000")

Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.