james.analysis: Analysis Tools for the 'JAMES' Framework.

In james.analysis: Analysis Tools for the 'JAMES' Framework

Description

This package can be used to further analyze and visualize results of studies performed with the analysis tools in 'JAMES', a modern object-oriented Java framework for discrete optimization using local search metaheuristics (see references). Functions are provided to plot convergence curves, draw box plots of solution quality or convergence times and to summarize, manipulate or extract data from the results.

Details

Package:	james.analysis
Type:	Package
Version:	1.0.1
Date:	2015-06-18
License:	MIT

Example data

james

Data manipulation functions

• readJAMES

• reduceJAMES

• mergeJAMES

• getProblems

• getSearches

• getSearchRuns

• getNumSearchRuns

• getBestSolutionValues

• getBestSolutions

• getConvergenceTimes

Plot functions

• plotConvergence

• boxplot.james

Author(s)

Herman De Beukelaer <[e-mail]>

References

'JAMES' Website: http://www.jamesframework.org

Examples

# load example data
data(james)
summary(james)

# plot convergence curves for coconut data set
plotConvergence(james, problem = "coconut", min.time = 1000, max.time = 100000)

# create box plots of solution values (quality) and convergence times
boxplot(james, problem = "coconut")
boxplot(james, problem = "coconut", type = "time")

# extract solution values and convergence times for parallel tempering and random descent
values.pt <- getBestSolutionValues(james, problem = "coconut", search = "Parallel Tempering")
times.pt <- getConvergenceTimes(james, problem = "coconut", search = "Parallel Tempering")
values.rd <- getBestSolutionValues(james, problem = "coconut", search = "Random Descent")
times.rd <- getConvergenceTimes(james, problem = "coconut", search = "Random Descent")

# perform wilcoxon test to compare distributions across algorithms
values.test <- wilcox.test(values.pt, values.rd)
values.test
times.test <- wilcox.test(times.pt, times.rd)
times.test

# adjust p-values for multiple testing
p.adjust(c(values.test$p.value, times.test$p.value))

james.analysis documentation built on May 1, 2019, 9:13 p.m.