modcmfitr (version 0.1.0)

mergeMultiplemCM: Merge multiple modified Connor-Mosimann distributions together

Description

This function merges the results of multiple experts' distributions using a numeric linear pool approach. It samples from the distributions of all experts individually many times (e.g. 100,000), then calculates the overall quantiles and medians from the samples. The function returns a matrix representing the lower, median and upper limits of the pooled distribution. This can then be fed into fitModCM() to generate a modified Connor-Mosimann distribution representing the overall spread of the experts' beliefs.

Usage

mergeMultiplemCM(NrSamples, RawData)

Arguments

NrSamples

Vector of length 1. Sets the number of samples to draw from each expert's mCM distribution

RawData

Data frame of mCM parameters. Must have six columns: expert, outcome, a, b, L, U of modified Connor-Mosimann distribution. Note last row of parameters will always be zero. Columns 1:6 of the output from function fitMultipleCM() are in the correct format for this.

Examples

# NOT RUN {
NrSamples <- 100000
RawData <- data.frame(expert = as.character(c(1,1,1,2,2,2)),
                      Outcome = as.factor(c("Remission","Progression","Dead",
                                           "Remission","Progression","Dead")),
                      a = as.numeric(c(6.0786, 0.2245, 0, 6.9214, 4.5259, 0)),
                      b = as.numeric(c(7.5900, 0.5866, 0, 1.7187, 3.1892, 0)),
                      L = as.numeric(c(0.3400, 0.4839, 0, 0.0152, 0.2390, 0)),
                      U = as.numeric(c(0.7917, 0.9213, 0, 0.7106, 0.9970, 0)))
mergeMultiplemCM(NrSamples,RawData)
# }