importance_sampling_from_stream: Runs Importance sampling evidence updating from an AMIDST...

In ramidst: An Interface to the AMIDST Toolbox for Data Stream Processing

Description

Runs Importance sampling evidence updating from an AMIDST data stream

Usage

importance_sampling_from_stream(network, target_variable, evidence_variables,
  input_stream, sample_size, parallel = T, seed = 3L)

Arguments

network	a java object of class BayesianNetwork over which the computations will be carried out
target_variable	a string representing the name of variable whose posterior distribution will be computed
evidence_variables	a vector with the names of the observed variables
input_stream	and AMIDST data stream
sample_size	the sample size to be used during the simulation
parallel	a boolean indicating whether or not the items in the sample will be generated in parallel (when allowed by the system)
seed	the seed for the genertion of random numbers

Value

a data.frame with the posterior distribution of the target variable for each item in the strea

Note

The function computes the posterior distribution given some evidence for all the items in the input stream.

References

A. Salmeron, D. Ramos-Lopez, H. Borchani, A.M. Martinez, A.R. Masegosa, A. Fernandez, H. Langseth, A.L. Madsen, T.D. Nielsen (2015) Parallel importance sampling in conditional linear Gaussian networks. CAEPIA'2015. Lecture Notes in Artificial Intelligence 9422, 36-46.

Examples

network <- load_amidst_bn(system.file("extdata","WasteIncinerator.bn",
package="ramidst"))
sample_stream <- amidst_data_stream(system.file("extdata",
"WasteIncineratorSample.arff",package="ramidst"))
posterior <- importance_sampling_from_stream(network,"B",c("F","E"),
sample_stream,50L)
posterior
posterior <- importance_sampling_from_stream(network,"L",c("F","E"),
sample_stream,50L)
posterior

ramidst documentation built on May 30, 2017, 5:27 a.m.