createTuningLinePlot: Create a tuning line plot

In collinleiber/benchmarkVis: Benchmark result visualisation

Description

This plot is just usable if the algorithm.parameter column contains an iteration value. Create a plotly line plot out of a benchmarkVis compatible data table. The created line plot shows the change of the specified measure for each iteration. A cumulative function can be defined to get a better feeling of the course of the line chart. Possible parameters for the cumulative function are: "id" (identity function), "min", "max", "mean" - default is min. If desired a histogram of the measure values can be plotted in the background to easily see the distribution in the data.table. Additionally a algorithm parameter can be defined to be plotted as circles on top of the line chart. This way the connection between measure result and parameter can be investigated. x-Axis: the iteration. y-Axis: the measure. y2-Axis (on the right): the optional algorithm parameter

Usage

createTuningLinePlot(dt, measure, cumulative.function = "min",
  show.histogram = FALSE, tuning.parameter = "none")

Arguments

dt	compatible data table
measure	the column name containing the results of a measure
cumulative.function	the cumulative function to use for the measure values (default: "min")
show.histogram	shows the histogram of the measure values in the background (default: FALSE)
tuning.parameter	the algorithm parameter (default: "none")

Value

a plotly line plot

Examples

createTuningLinePlot(mlr.tuning.example, "measure.acc.test.mean", "id", TRUE, "C")

collinleiber/benchmarkVis documentation built on May 28, 2019, 8:23 a.m.