ncdDetectOverdispersion: Add discrete stochastic random variables of arbitrary...

In MaleneJuul/ncdDetectTools: Functions to perform convolution

Description

Add discrete stochastic random variables of arbitrary dimensions with overdispersion

Usage

ncdDetectOverdispersion(predictions, scores, overdispersion = 0, N = 10,
  method = c("naive", "is", "numeric"), observations = NA, thres = NA)

Arguments

overdispersion	The standard deviation of a random effect in logit space. The random effect is not added to the first column.
dat	A data.table with columns 'x', 'y', and 'probabilities'. Column 'x' is a numeric indicator unique for each random variable to be convoluted together. Column 'y' is the outcome value, and colum 'probabilities' contains the probability corresponding to each value of the 'y' column.
threshold	An optional number to set a threshold. If set, the probabilities of all possible outcomes of the sum of the random variables are only caluclated below this value.
verbose	A logical indicator which defaults to FALSE. Set to TRUE to print information.

Value

the convoluted distribution.

Examples

convolution(dat = data.table(x = c(rep(1, 2), rep(2, 3), rep(3, 4)), y = c(1:2, 1:3, 1:4), 
    probability = c(3/4, 1/4, rep(1/3, 3), (1:4)/10)))
convolution(dat = data.table(x = c(rep(1, 2), rep(2, 3), rep(3, 4)), y = c(1:2, 1:3, 1:4), 
    probability = c(3/4, 1/4, rep(1/3, 3), (1:4)/10)), threshold = 6)

MaleneJuul/ncdDetectTools documentation built on May 8, 2019, 3:24 p.m.