arulesNBMiner (version 0.1-8)

NBMinerParameters: Estimate Global Model Parameters from Data

Description

Estimate the global negative binomial data model used by the NBMiner and create an appropriate parameter object.

Usage

NBMinerParameters(data, trim = 0.01, pi = 0.99, theta = 0.5,
    minlen = 1, maxlen = 5, rules = FALSE,
    plot = FALSE, verbose = FALSE, getdata = FALSE)

Arguments

data

the data as a object of class transactions.

trim

fraction of incidences to trim off the tail of the frequency distribution of the data.

pi

precision threshold \(\pi\).

theta

pruning parameter \(\theta\).

minlen

minimum number of items in found itemsets (default: 1).

maxlen

maximal number of items in found itemsets (default: 5).

rules

mine NB-precise rules instead of NB-frequent itemsets?

plot

plot the model?

verbose

use verbose output for the estimation procedure.

getdata

get also the observed and estimated counts.

Value

an object of class NBMinerParameter for NBMiner.

Details

Uses the EM algorithm to estimate the global NB model for the data. The EM algorithm is used since the zero class (items which do not occur in the dataset) is not included in the data. The result are the two NB parameters \(k\) and \(a\), where \(a\) is rescaled by dividing it by the number of incidences in the data (this is needed by the NBMiner). Also the real number of items \(n\) is a result of the estimation.

theta and pi are just taken and added to the resulting parameter object.

References

Michael Hahsler. A model-based frequency constraint for mining associations from transaction data. Data Mining and Knowledge Discovery,13(2):137-166, September 2006.

See Also

NBMiner, transactions-class

Examples

# NOT RUN {
data("Epub")

param <- NBMinerParameters(Epub, trim = 0.05, plot = TRUE, verbose = TRUE)
param
# }