[Bug]: Machine Learning Clustering Neighborhood based - Cluster Means plot mixes up values

[zander86]

JASP Version

0.19.3

Commit ID

No response

JASP Module

Machine Learning

What analysis are you seeing the problem on?

Neighborhood-Based Clustering

What OS are you seeing the problem on?

Windows 11

Bug Description

The Cluster means plot is mixing up the values from 2 variables. I cannot better explain it than snown in the screenshot below. All the plots are correct, only the table mixes up Sunlight and Rainfall in this case.
printscreen:

Jasp file

plant_dataset.zip

Expected Behaviour

Just mix up the column headers rainfall and Sunlight.

Steps to Reproduce

neigborhood based clustering.
Features: Sunlight and reainfalll
tick cluster means and cluster means barplot.
check the differences.

Log (if any)

No response

More Debug Information

No response

Final Checklist

• I have included a screenshot showcasing the issue, if possible.
• I have included a JASP file (zipped) or data file that causes the crash/bug, if applicable.
• I have accurately described the bug, and steps to reproduce it.

[koenderks]

@vandenman I cannot figure out why this occurs in JASP. Other datasets work fine.

I tried to reproduce this in R using :

options <- jaspTools::analysisOptions('
                                      {
	"name" : "mlClusteringKMeans",
	"options" : 
	{
		"addEllips" : false,
		"addPredictions" : true,
		"algorithm" : "Lloyd",
		"centers" : "means",
		"clusterDensityPlot" : true,
		"clusterDensityPlotSingleFigure" : true,
		"clusterMeanPlot" : true,
		"clusterMeanPlotBarPlot" : true,
		"clusterMeanPlotSingleFigure" : true,
		"distance" : "euclidean",
		"elbowMethodPlot" : false,
		"manualNumberOfClusters" : 3,
		"matrixPlot" : true,
		"maxNumberIterations" : 25,
		"maxNumberOfClusters" : 10,
		"modelOptimization" : "manual",
		"modelOptimizationMethod" : "bic",
		"noOfRandomSets" : 25,
		"plotHeight" : 320,
		"plotWidth" : 480,
		"predictionsColumn" : "Predictions",
		"predictors" : 
			[
				"Sunlight..hours.day.",
				"Rainfall..mm.year."
			],
		"saveModel" : false,
		"savePath" : "",
		"scaleVariables" : true,
		"seed" : 1,
		"setSeed" : true,
		"tableClusterInformation" : true,
		"tableClusterInformationBetweenSumOfSquares" : false,
		"tableClusterInformationCentroids" : true,
		"tableClusterInformationSilhouetteScore" : true,
		"tableClusterInformationTotalSumOfSquares" : false,
		"tableClusterInformationWithinSumOfSquares" : true,
		"tableClusterMeans" : true,
		"tsneClusterPlot" : false,
		"tsneClusterPlotLabels" : false,
		"tsneClusterPlotLegend" : true,
		"validationMeasures" : false
	},
	"perform" : "run",
	"preloadData" : true,
	"resultFont" : "Lucida Grande",
	"revision" : 5,
	"rfile" : "",
	"title" : "Neighborhood-Based Clustering",
	"typeRequest" : "analysis"
}
                                      ')
jaspTools::runAnalysis("mlClusteringKMeans", dataset = read.csv("~/Downloads/plant_dataset.csv", sep = ";"), options)

With dataset: plant_dataset.csv

but there I just get matching plots (different from JASP)...

How can my R output be different from the JASP results?

JASP file: plant_dataset.jasp.zip

[koenderks]

Another example in which the R output is correct but JASP somehow gives different output with the exact same options:

Random forest clustering:

R:

options <- jaspTools::analysisOptions('
	{
	"name" : "mlClusteringRandomForest",
	"options" : 
	{
		"addEllips" : false,
		"addPredictions" : false,
		"clusterDensityPlot" : false,
		"clusterDensityPlotSingleFigure" : true,
		"clusterMeanPlot" : true,
		"clusterMeanPlotBarPlot" : true,
		"clusterMeanPlotSingleFigure" : true,
		"elbowMethodPlot" : false,
		"featureImportancePermutations" : 50,
		"featureImportanceTable" : false,
		"manualNumberOfClusters" : 3,
		"matrixPlot" : true,
		"maxNumberOfClusters" : 10,
		"modelOptimization" : "manual",
		"modelOptimizationMethod" : "bic",
		"numberOfTrees" : 1000,
		"plotHeight" : 320,
		"plotWidth" : 480,
		"predictionsColumn" : "",
		"predictors" : 
			[
				"Sunlight..hours.day.",
				"Rainfall..mm.year."
			],
		"saveModel" : false,
		"savePath" : "",
		"scaleVariables" : true,
		"seed" : 1,
		"setSeed" : true,
		"tableClusterInformation" : true,
		"tableClusterInformationBetweenSumOfSquares" : false,
		"tableClusterInformationCentroids" : false,
		"tableClusterInformationSilhouetteScore" : false,
		"tableClusterInformationTotalSumOfSquares" : false,
		"tableClusterInformationWithinSumOfSquares" : true,
		"tableClusterMeans" : false,
		"tsneClusterPlot" : false,
		"tsneClusterPlotLabels" : false,
		"tsneClusterPlotLegend" : true,
		"validationMeasures" : false
	},
	"perform" : "run",
	"preloadData" : true,
	"resultFont" : "Lucida Grande",
	"revision" : 1,
	"rfile" : "",
	"title" : "Random Forest Clustering",
	"typeRequest" : "analysis"
}
                                      ')
jaspTools::runAnalysis("mlClusteringRandomForest", dataset = read.csv("~/Downloads/plant_dataset.csv", sep = ";"), options)

Gives:

While JASP file plant_dataset.jasp.zip with exact same options gives:

How can this be so different?

[koenderks]

It also seems specific to this dataset, as if I try this with the iris dataset, I get the same results between R and JASP:

options <- jaspTools::analysisOptions('
	{
	"name" : "mlClusteringRandomForest",
	"options" : 
	{
		"addEllips" : false,
		"addPredictions" : false,
		"clusterDensityPlot" : false,
		"clusterDensityPlotSingleFigure" : true,
		"clusterMeanPlot" : true,
		"clusterMeanPlotBarPlot" : true,
		"clusterMeanPlotSingleFigure" : true,
		"elbowMethodPlot" : false,
		"featureImportancePermutations" : 50,
		"featureImportanceTable" : false,
		"manualNumberOfClusters" : 3,
		"matrixPlot" : true,
		"maxNumberOfClusters" : 10,
		"modelOptimization" : "manual",
		"modelOptimizationMethod" : "bic",
		"numberOfTrees" : 1000,
		"plotHeight" : 320,
		"plotWidth" : 480,
		"predictionsColumn" : "",
		"predictors" : 
			[
				"Sepal.Length",
				"Sepal.Width"
			],
		"saveModel" : false,
		"savePath" : "",
		"scaleVariables" : true,
		"seed" : 1,
		"setSeed" : true,
		"tableClusterInformation" : true,
		"tableClusterInformationBetweenSumOfSquares" : false,
		"tableClusterInformationCentroids" : false,
		"tableClusterInformationSilhouetteScore" : false,
		"tableClusterInformationTotalSumOfSquares" : false,
		"tableClusterInformationWithinSumOfSquares" : true,
		"tableClusterMeans" : true,
		"tsneClusterPlot" : false,
		"tsneClusterPlotLabels" : false,
		"tsneClusterPlotLegend" : true,
		"validationMeasures" : false
	},
	"perform" : "run",
	"preloadData" : true,
	"resultFont" : "Lucida Grande",
	"revision" : 1,
	"rfile" : "",
	"title" : "Random Forest Clustering",
	"typeRequest" : "analysis"
}
                                      ')
jaspTools::runAnalysis("mlClusteringRandomForest", dataset = "iris.csv", options)

Gives:

And JASP (iris.jasp.zip) gives: