ipptoolbox (version 1.2)

aggregation: Aggregation

Description

Various functions to aggregate a set of estimated BPAs to a single aggregation.

Usage

dsdempstersrule(..., maxfocals = 1e+07)
dsenvelope(...)
dsintersect(...)
dssaveintersect(..., method = "averaging")
dsxaveraging(..., w = NULL, maxfocals = 1e+07)
dsxenvelope(..., maxfocals = 1e+07)
dswavg(..., w = NULL)
dswmix(..., w = NULL)

Arguments

…

1...n BPAs to be aggregated.

w

Vector of n weights for weighted aggregations

maxfocals

Maximal number of focal elements of the result. Warning: If not constrained, some rules use time and memory \(O(|m|^|x|)\), |m| number of focals by structure, |x| number of structures. Meaning: 5 structures with 1000 focals: Maximal \(10^15\) bytes

method

method="averaging": In case of empty intersection, focal elements will be averaged. method="averaging": In case of empty intersection, focal elements will be enveloped.

Value

Aggregation of the BPAs according to the selected rule.

Details

A set of functions to aggregate a set of estimated BPAs to a single aggregation. Some functions have an optional maxfocals parameter that limits the size of the resulting BPA (which theoretically is the product of all focal elements). The vector w contains weights in case of a weighted aggregation type. The dssaveintersect method has an optional parameter that specifies the alternative rule in case of empty intersections.

In the example, two p-boxes defined from exponential distributions are aggregated using dsdempstersrule and dswmix.

References

Sentz, K. and S. Ferson (2002). Combination of Evidence in Dempster-Shafer Theory. Sandia Report. Albuquerque, USA, Sandia National Laboratories.

Examples

# NOT RUN {
lambda1=dsstruct(c(2,3,1))
dss1=dsodf('qexp',100,lambda1);
lambda2=dsstruct(c(5,6,1))
dss2=dsodf('qexp',100,lambda2);
y=dsdempstersrule(dss1,dss2);
y2=dswmix(dss1,dss2,w=c(2,0.5));
dscdf(y);

dscdf(y2);
# }