probInterval: Various method around IPIntervals
In imptree: Classification Trees with Imprecise Probabilities

Description Usage Arguments Value Author(s) See Also Examples

Calculation of probability intervals, and their maximal and minimal entropy

1
2
3

probInterval(table, iptype = c("IDM", "NPI", "NPIapprox"),
  entropymin = TRUE, entropymax = TRUE, correction = c("no",
  "strobl", "abellan"), s = 1)

`table`	integer vector of absolute frequencies
`iptype`	method for calculating the probability intervals of `table`. `"IDM"` for the imprecise Dirichlet model (default), `"NPI"` for use of the nonparametric predictive inference approach and `"NPIapprox"` for use of the approximate algorithm obtaining maximal entropy of NPI generated probability intervals.
`entropymin`	Calculation of one distribution with minimal entropy, including the actual value of the minimal entropy (default: `TRUE`)
`entropymax`	Calculation of the distribution with maximal entropy, including the actual value of the maximal entropy (default: `TRUE`)
`correction`	Entropy correction to be carried out, ignorned if `(entropymin \|\| entropymax) == FALSE` (default `"no"`), see `imptree_params`
`s`	Hyperparamter of the IDM (`s >= 0`), see `imptree_params` (ignored for `iptype == "NPI"`)

A list with 5 named entries:

`probint`	matrix with 3 rows and `length(table)` columns: in the rows are the abosulte frequencies, the lower bound (`"lower"`) and the upper bound (`"upper"`) of the event-wise probabilities.
`maxEntDist`	The (unique) probability distribution with maximal entropy
`maxEntCorr`	The value of the (corrected) maximal entropy
`minEntDist`	A probability distribution with minimal entropy, as it is not necessarily unqiue there may be others
`minEntCorr`	The value of the (corrected) minimal entropy

Paul Fink Paul.Fink@stat.uni-muenchen.de

imptree_params

## Artificial vector of absolute frequencies
obs <- c(a = 1,b = 2, c = 10, d = 30, e = 5)

## probability interval by NPI, including only information on the
## mininum entropy distribution, using no entropy correction
probInterval(obs, iptype = "NPI", entropymax = FALSE)

## probability interval by IDM, including information on the
## minimum and maximum entropy distribution with s = 2 and correction
## according to 'strobl'
probInterval(obs, iptype = "IDM", correction = "strobl", s = 2)