james.analysis (version 1.0.1)

james: Results of example algorithm comparison

Description

Contains results of an example analysis performed with the 'JAMES' extensions module. The performance of two algorithms is compared (random descent and parallel tempering) for a core selection problem in which the mean entry-to-nearest-entry distance is maximized. Four different data sets have been analyzed. Details about the performed analysis are provided at the website (see below).

Usage

james

Format

S3 object of class "james", as if produced by readJAMES.

Source

http://www.jamesframework.org/examples/#analysis

See Also

readJAMES

Examples

# load 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))