Bugfixxing (9a1ab1ca) · Commits · Jakub Peschel / ADAMiSS Sequence Experiment

src/main/java/cz/muni/disa/adamiss/se/stats/objects/ExperimentReport.java

+6 −5

Original line number	Original line	Diff line number	Diff line
	@@ -71,6 +71,7 @@ public record ExperimentReport(String name, DBReport dbStats, int threshold, dou
	avgRepeatedMap.get(algo).add(new Pair<>(report.dbStats().avgRepeatedItems(), report.time()));		avgRepeatedMap.get(algo).add(new Pair<>(report.dbStats().avgRepeatedItems(), report.time()));
	thresholdMap.get(algo).add(new Pair<>(report.threshold(), report.time()));		thresholdMap.get(algo).add(new Pair<>(report.threshold(), report.time()));
	}		}
			}
	mapSort(sizeMap);		mapSort(sizeMap);
	mapSort(uniqueMap);		mapSort(uniqueMap);
	mapSort(minLengthMap);		mapSort(minLengthMap);
	@@ -93,11 +94,10 @@ public record ExperimentReport(String name, DBReport dbStats, int threshold, dou
	mapPrint("avgRepeated", avgRepeatedMap);		mapPrint("avgRepeated", avgRepeatedMap);
	mapPrint("threshold", thresholdMap);		mapPrint("threshold", thresholdMap);
	}		}
	}

	private static <K extends Comparable,V> void mapSort(Map<Algo, List<Pair<K,V>>> map){		private static <K extends Comparable,V> void mapSort(Map<Algo, List<Pair<K,V>>> map){
	for (Algo algo : map.keySet()){		for (Algo algo : map.keySet()){
	map.get(algo).stream().sorted(Comparator.comparing(Pair::getKey));		map.get(algo).sort(Comparator.comparing(Pair::getKey));
	}		}
	}		}
	private static <K extends Comparable,V extends Comparable> void mapPrint(String name, Map<Algo, List<Pair<K,V>>> map){		private static <K extends Comparable,V extends Comparable> void mapPrint(String name, Map<Algo, List<Pair<K,V>>> map){
	@@ -115,8 +115,9 @@ public record ExperimentReport(String name, DBReport dbStats, int threshold, dou
	for (var p : a)		for (var p : a)
	uniqueK.add(p.getKey());		uniqueK.add(p.getKey());
	}		}
			List<K> listUniqueK = new ArrayList<>(uniqueK);
	for (K key: uniqueK) {		listUniqueK.sort(K::compareTo);
			for (K key: listUniqueK) {
	transposedMap.put(key, new ArrayList<>());		transposedMap.put(key, new ArrayList<>());
	for(Algo algo : algoOrder){		for(Algo algo : algoOrder){

	@@ -137,7 +138,7 @@ public record ExperimentReport(String name, DBReport dbStats, int threshold, dou
	List<K> keyOrder = new ArrayList<>(transposedMap.keySet());		List<K> keyOrder = new ArrayList<>(transposedMap.keySet());
	keyOrder = keyOrder.stream().sorted().collect(Collectors.toList());		keyOrder = keyOrder.stream().sorted().collect(Collectors.toList());
	for (K key : keyOrder){		for (K key : keyOrder){
	valuesCSV.append(key.toString());		valuesCSV.append(key.toString()).append(", ");
	var pairList = transposedMap.get(key);		var pairList = transposedMap.get(key);
	for (var algo : algoOrder){		for (var algo : algoOrder){
	//todo:compute candle value and average		//todo:compute candle value and average

tools/experimentStats.p

+289 −22

Original line number	Original line	Diff line number	Diff line
	@@ -19,40 +19,307 @@ set style line 3 \
	linetype 1 linewidth 2 \		linetype 1 linewidth 2 \
	pointtype 7 pointsize 1.5		pointtype 7 pointsize 1.5

	set boxwidth 0.1 relative		set style line 4 \
			linecolor rgb '#00b300' \
			linetype 1 linewidth 2 \
			pointtype 7 pointsize 1.5

			set style line 5 \
			linecolor rgb '#0000cc' \
			linetype 1 linewidth 2 \
			pointtype 7 pointsize 1.5

			set style line 6 \
			linecolor rgb '#cc0000' \
			linetype 1 linewidth 2 \
			pointtype 7 pointsize 1.5

	set style fill empty		set style fill empty

	set output 'sizeExp.png'		set output 'sizeExp.png'
			stats 'size.csv' using 3 name 'MIN_GSP' nooutput
			stats 'size.csv' using 5 name 'MAX_GSP' nooutput
			stats 'size.csv' using 7 name 'MIN_PS' nooutput
			stats 'size.csv' using 10 name 'MAX_PS' nooutput
			stats 'size.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'size.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

			stats 'size.csv' using 1 name 'X' nooutput
			boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
	plot 'size.csv' using 1:4 with lp title 'GSP' ls 1 , \		plot 'size.csv' using 1:4 with lp title 'GSP' ls 1 , \
	'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \		'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
	'' using 1:14 with lp title 'SPAM' ls 3, \		'' using 1:14 with lp title 'SPAM' ls 3, \
	'' using ($1-1):3:2:6:5 with candlesticks title 'GSP' whiskerbars, \		'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
	'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars, \		'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
	'' using ($1+1):13:12:16:15 with candlesticks title 'SPAM' whiskerbars		'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

	#set output 'unique.png'		set xrange [:]
	#plot 'unique.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		set yrange [:]
			set output 'unique.png'
			stats 'unique.csv' using 3 name 'MIN_GSP' nooutput
			stats 'unique.csv' using 5 name 'MAX_GSP' nooutput
			stats 'unique.csv' using 7 name 'MIN_PS' nooutput
			stats 'unique.csv' using 10 name 'MAX_PS' nooutput
			stats 'unique.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'unique.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

	#set output 'minLength.png'		stats 'unique.csv' using 1 name 'X' nooutput
	#plot 'minLength.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'unique.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

	#set output 'avgLength.png'		set xrange [:]
	#plot 'avgLength.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		set yrange [:]
			set output 'minLength.png'
			stats 'minLength.csv' using 3 name 'MIN_GSP' nooutput
			stats 'minLength.csv' using 5 name 'MAX_GSP' nooutput
			stats 'minLength.csv' using 7 name 'MIN_PS' nooutput
			stats 'minLength.csv' using 10 name 'MAX_PS' nooutput
			stats 'minLength.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'minLength.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

	#set output 'maxLength.png'		stats 'minLength.csv' using 1 name 'X' nooutput
	#plot 'maxLength.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'minLength.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

	#set output 'minRepeatibility.png'		set xrange [:]
	#plot 'minRepeatibility.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		set yrange [:]
			set output 'avgLength.png'
			stats 'avgLength.csv' using 3 name 'MIN_GSP' nooutput
			stats 'avgLength.csv' using 5 name 'MAX_GSP' nooutput
			stats 'avgLength.csv' using 7 name 'MIN_PS' nooutput
			stats 'avgLength.csv' using 10 name 'MAX_PS' nooutput
			stats 'avgLength.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'avgLength.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

	#set output 'avgRepeatibility.png'		stats 'avgLength.csv' using 1 name 'X' nooutput
	#plot 'avgRepeatibility.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'avgLength.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

			set xrange [:]
			set yrange [:]
			set output 'maxLength.png'
			stats 'maxLength.csv' using 3 name 'MIN_GSP' nooutput
			stats 'maxLength.csv' using 5 name 'MAX_GSP' nooutput
			stats 'maxLength.csv' using 7 name 'MIN_PS' nooutput
			stats 'maxLength.csv' using 10 name 'MAX_PS' nooutput
			stats 'maxLength.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'maxLength.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

	#set output 'maxRepeatibility.png'		stats 'maxLength.csv' using 1 name 'X' nooutput
	#plot 'maxRepeatibility.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'maxLength.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

	#set output 'avgRepeated.png'		set xrange [:]
	#plot 'avgRepeated.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		set yrange [:]
			set output 'minRepeatibility.png'
			stats 'minRepeatibility.csv' using 3 name 'MIN_GSP' nooutput
			stats 'minRepeatibility.csv' using 5 name 'MAX_GSP' nooutput
			stats 'minRepeatibility.csv' using 7 name 'MIN_PS' nooutput
			stats 'minRepeatibility.csv' using 10 name 'MAX_PS' nooutput
			stats 'minRepeatibility.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'minRepeatibility.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

	#set output 'threshold.png'		stats 'minRepeatibility.csv' using 1 name 'X' nooutput
	#plot 'threshold.csv' using 1:2 with lp ls 1, '' using 1:3 with lp ls 2, '' using 1:4 with lp ls 3		boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			#plot 'minRepeatibility.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

			set xrange [:]
			set yrange [:]
			set output 'avgRepeatibility.png'
			stats 'avgRepeatibility.csv' using 3 name 'MIN_GSP' nooutput
			stats 'avgRepeatibility.csv' using 5 name 'MAX_GSP' nooutput
			stats 'avgRepeatibility.csv' using 7 name 'MIN_PS' nooutput
			stats 'avgRepeatibility.csv' using 10 name 'MAX_PS' nooutput
			stats 'avgRepeatibility.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'avgRepeatibility.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

			stats 'avgRepeatibility.csv' using 1 name 'X' nooutput
			boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'avgRepeatibility.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

			set xrange [:]
			set yrange [:]
			set output 'maxRepeatibility.png'
			stats 'maxRepeatibility.csv' using 3 name 'MIN_GSP' nooutput
			stats 'maxRepeatibility.csv' using 5 name 'MAX_GSP' nooutput
			stats 'maxRepeatibility.csv' using 7 name 'MIN_PS' nooutput
			stats 'maxRepeatibility.csv' using 10 name 'MAX_PS' nooutput
			stats 'maxRepeatibility.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'maxRepeatibility.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

			stats 'maxRepeatibility.csv' using 1 name 'X' nooutput
			boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'maxRepeatibility.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

			set xrange [:]
			set yrange [:]
			set output 'avgRepeated.png'
			stats 'avgRepeated.csv' using 3 name 'MIN_GSP' nooutput
			stats 'avgRepeated.csv' using 5 name 'MAX_GSP' nooutput
			stats 'avgRepeated.csv' using 7 name 'MIN_PS' nooutput
			stats 'avgRepeated.csv' using 10 name 'MAX_PS' nooutput
			stats 'avgRepeated.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'avgRepeated.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

			stats 'avgRepeated.csv' using 1 name 'X' nooutput
			boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'avgRepeated.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6

			set xrange [:]
			set yrange [:]
			set output 'threshold.png'
			stats 'threshold.csv' using 3 name 'MIN_GSP' nooutput
			stats 'threshold.csv' using 5 name 'MAX_GSP' nooutput
			stats 'threshold.csv' using 7 name 'MIN_PS' nooutput
			stats 'threshold.csv' using 10 name 'MAX_PS' nooutput
			stats 'threshold.csv' using 13 name 'MIN_SPAM' nooutput
			stats 'threshold.csv' using 15 name 'MAX_SPAM' nooutput
			MAX_max = (MAX_GSP_max + MAX_PS_max + MAX_SPAM_max)/3
			MIN_min = (MIN_GSP_min + MIN_PS_min + MIN_SPAM_min)/3
			rangeOffset = ((MAX_max - MIN_min)/10)

			stats 'threshold.csv' using 1 name 'X' nooutput
			boxes = X_records*3
			rangesize = X_max- X_min
			rangeoffset = rangesize / 10
			boxsize = (rangesize / boxes)/10
			print boxsize
			set boxwidth boxsize
			set xrange [(X_min - rangeoffset):(X_max + rangeoffset)]
			set yrange [(MIN_min-rangeOffset):(MAX_max+rangeOffset)]
			plot 'threshold.csv' using 1:4 with lp title 'GSP' ls 1 , \
			'' using 1:9 with lp title 'PREFIX_SPAN' ls 2 , \
			'' using 1:14 with lp title 'SPAM' ls 3, \
			'' using ($1- (boxsize)):3:2:6:5 with candlesticks title 'GSP' whiskerbars ls 4, \
			'' using ($1):8:7:11:10 with candlesticks title 'PREFIX_SPAN' whiskerbars ls 5, \
			'' using ($1+ (boxsize)):13:12:16:15 with candlesticks title 'SPAM' whiskerbars ls 6