add percentile information if available

This commit is contained in:
2021-04-02 18:29:18 +02:00
parent f56744eb6b
commit a1b4c7006d
15 changed files with 322 additions and 168 deletions

View File

@@ -19,6 +19,7 @@ import { MatFormFieldModule } from '@angular/material/form-field';
import { MatInputModule } from '@angular/material/input';
import {MatProgressBarModule} from '@angular/material/progress-bar';
import {MatProgressSpinnerModule} from '@angular/material/progress-spinner';
import {MatRadioModule} from '@angular/material/radio';
import {MatSnackBarModule} from '@angular/material/snack-bar';
import {MatTooltipModule} from '@angular/material/tooltip';
import { YAxisDefinitionComponent } from './y-axis-definition/y-axis-definition.component';
@@ -56,6 +57,7 @@ import { ImageToggleComponent } from './image-toggle/image-toggle.component';
MatCheckboxModule,
MatFormFieldModule,
MatInputModule,
MatRadioModule,
MatProgressBarModule,
MatProgressSpinnerModule,
MatSelectModule,

View File

@@ -1,18 +1,29 @@
<mat-radio-group [(ngModel)]="valueFormat" aria-label="Value format">
<mat-radio-button value="time">Time</mat-radio-button>
<mat-radio-button value="numbers">Numbers</mat-radio-button>
</mat-radio-group>
<table class="gallery-item-details">
<tr>
<th>Name</th>
<th>Type</th>
<th>Values</th>
<th>Avg</th>
<td *ngFor="let label of percentilesToPlot.keys()">{{label}}</td>
<td>Max</td>
</tr>
<tr *ngFor="let stat of stats.dataSeriesStats">
<td>{{ stat.name }}</td>
<td><div class="{{ pointTypeClass(stat.dashTypeAndColor) }}" title="{{ stat.name }}"></div></td>
<td>{{ stat.values }}</td>
<td>{{ utils.formatMs(stat.average) }}</td>
<td>{{ utils.format(stat.average, valueFormat) }}</td>
<td *ngFor="let key of percentilesToPlot.keys()">{{utils.format(stat.percentiles[percentilesToPlot.get(key)], valueFormat)}}</td>
<td>{{ utils.format(stat.maxValue, valueFormat)}}</td>
</tr>
</table>
<table class="gallery-item-details-matrix">
<div *ngIf="stats.dataSeriesStats.length > 1">
<h2>Compare Averages</h2>
<table class="gallery-item-details-matrix">
<tr>
<th></th>
<th *ngFor="let statsCol of stats.dataSeriesStats">
@@ -25,4 +36,24 @@
{{ utils.toPercent(statsRow.average / statsCol.average) }}
</td>
</tr>
</table>
</table>
<h2>Compare Percentiles</h2>
<div *ngFor="let p of percentilesToPlot.keys()">
<h3>{{p}} percentile</h3>
<table class="gallery-item-details-matrix">
<tr>
<th></th>
<th *ngFor="let statsCol of stats.dataSeriesStats">
<div class="{{ pointTypeClass(statsCol.dashTypeAndColor) }}" title="{{ statsCol.name }}"></div>
</th>
</tr>
<tr *ngFor="let statsRow of stats.dataSeriesStats">
<td><div class="{{ pointTypeClass(statsRow.dashTypeAndColor) }}" title="{{ statsRow.name }}"></div></td>
<td *ngFor="let statsCol of stats.dataSeriesStats">
{{ utils.toPercent(statsRow.percentiles[percentilesToPlot.get(p)] / statsCol.percentiles[percentilesToPlot.get(p)]) }}
</td>
</tr>
</table>
</div>
</div>

View File

@@ -28,3 +28,8 @@
.plot-details-plotType_e51e10 {background-position-y: -40px;}
.plot-details-plotType_57a1c2 {background-position-y: -48px;}
.plot-details-plotType_bd36c2 {background-position-y: -56px;}
.gallery-item-details td {
white-space: pre;
}

View File

@@ -12,10 +12,38 @@ export class PlotDetailsComponent {
@Input()
stats: PlotResponseStats;
hasPercentiles = false;
valueFormat = "time";
percentilesToPlot : Map<string,string> = new Map();
constructor(public utils: UtilService){
}
ngOnInit() {
this.hasPercentiles = false;
this.percentilesToPlot.clear();
for (let i = 0; i < this.stats.dataSeriesStats.length; i++)
{
const stat = this.stats.dataSeriesStats[i];
if (stat.percentiles.hasOwnProperty("50.000"))
{
this.hasPercentiles = true;
this.percentilesToPlot.set('median','50.000');
this.percentilesToPlot.set('75th','75.000');
this.percentilesToPlot.set('95th','95.000');
this.percentilesToPlot.set('99th','99.000');
break;
}
}
}
percentile(value: number): string {
return this.utils.format(value, this.valueFormat);
}
pointTypeClass(typeAndColor: DashTypeAndColor): string {
return "plot-details-plotType"
+" plot-details-plotType_"+typeAndColor.pointType

View File

@@ -19,4 +19,5 @@ img {
bottom: 5px;
background-color: white;
box-shadow: 5px 5px 10px 0px #e0e0e0;
overflow: auto;
}

View File

@@ -249,6 +249,7 @@ export class DataSeriesStats {
average : number;
plottedValues : number;
dashTypeAndColor: DashTypeAndColor;
percentiles: Map<string, number>
}
export class DashTypeAndColor {

View File

@@ -9,6 +9,14 @@ export class UtilService {
constructor() {
}
format(value: number, type: string) {
if (type == "time"){
return this.formatMs(value);
} else {
return ""+value;
}
}
formatMs(valueInMs):string {
const ms = Math.floor(valueInMs % 1000);
const s = Math.floor((valueInMs / 1000) % 60);

View File

@@ -48,6 +48,7 @@
#filters {
grid-area: filters;
overflow: auto;
}
#filterpanel {
background-color: #f8f8f8;/*#fafafa;*/

View File

@@ -3,6 +3,7 @@ package org.lucares.pdb.plot.api;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.Objects;
import java.util.Optional;
import org.lucares.pdb.api.Tags;
@@ -27,4 +28,15 @@ public class AggregatorCollection {
return Optional.ofNullable(aggregators.get(type));
}
@SuppressWarnings("unchecked")
public <T extends CustomAggregator> Optional<T> getAggregator(final Class<T> aggregatorType) {
for (final CustomAggregator aggregator : aggregators.values()) {
if (Objects.equals(aggregator.getClass(), aggregatorType)) {
return Optional.of((T) aggregator);
}
}
return Optional.empty();
}
}

View File

@@ -8,7 +8,7 @@ import java.io.OutputStreamWriter;
import java.io.Writer;
import java.nio.charset.StandardCharsets;
import java.nio.file.Path;
import java.util.LinkedHashMap;
import java.util.Locale;
import org.lucares.collections.LongLongConsumer;
import org.lucares.collections.LongLongHashMap;
@@ -24,7 +24,7 @@ public class CumulativeDistributionCustomAggregator implements CustomAggregator
private long maxValue = 0;
private final LinkedHashMap<Double, Long> percentiles = new LinkedHashMap<>(POINTS);
private final Percentiles percentiles = new Percentiles(POINTS);
private final double stepSize;
@@ -49,7 +49,8 @@ public class CumulativeDistributionCustomAggregator implements CustomAggregator
if (newPercentile >= nextPercentile) {
double currentPercentile = lastPercentile + stepSize;
while (currentPercentile <= newPercentile) {
percentiles.put(currentPercentile, duration);
final String percentile = String.format(Locale.US, "%.3f", currentPercentile);
percentiles.put(percentile, duration);
currentPercentile += stepSize;
}
nextPercentile = currentPercentile;
@@ -61,10 +62,15 @@ public class CumulativeDistributionCustomAggregator implements CustomAggregator
return maxValue;
}
public LinkedHashMap<Double, Long> getPercentiles() {
public Percentiles getPercentiles() {
return percentiles;
}
public void collect(final LongLongHashMap map) {
map.forEachOrdered(this);
percentiles.put("100.000", maxValue);
}
}
// the rather large initial capacity should prevent too many grow&re-hash phases
@@ -84,14 +90,27 @@ public class CumulativeDistributionCustomAggregator implements CustomAggregator
totalValues++;
}
public Percentiles getPercentiles() {
final long start = System.nanoTime();
final ToPercentiles toPercentiles = new ToPercentiles(totalValues);
toPercentiles.collect(map);
final Percentiles result = toPercentiles.getPercentiles();
System.out.println("getPercentiles took: " + (System.nanoTime() - start) / 1_000_000.0 + " ms");
return result;
}
@Override
public AggregatedData getAggregatedData() {
try {
final char separator = ',';
final char newline = '\n';
final long start = System.nanoTime();
final ToPercentiles toPercentiles = new ToPercentiles(totalValues);
map.forEachOrdered(toPercentiles);
toPercentiles.collect(map);
System.out.println("getAggregated took: " + (System.nanoTime() - start) / 1_000_000.0 + " ms");
final File dataFile = File.createTempFile("data", ".dat", tmpDir.toFile());
try (final Writer output = new BufferedWriter(
@@ -107,12 +126,6 @@ public class CumulativeDistributionCustomAggregator implements CustomAggregator
data.append(value);
data.append(newline);
});
final long maxValue = toPercentiles.getMaxValue();
data.append(100);
data.append(separator);
data.append(maxValue);
data.append(newline);
}
output.write(data.toString());

View File

@@ -0,0 +1,32 @@
package org.lucares.pdb.plot.api;
import java.util.LinkedHashMap;
import java.util.Map;
/**
* Maps percentiles to their value. E.g.
*
* <pre>
* {"50.00": 123, "75.00": 567}
* </pre>
*
* This class uses Strings for the precentiles instead of doubles, because
* doubles are bad keys for maps.
*/
public class Percentiles extends LinkedHashMap<String, Long> {
private static final long serialVersionUID = 4957667781086113971L;
public Percentiles() {
super(0);
}
public Percentiles(final int initialSize) {
super(initialSize);
}
public Percentiles(final Map<String, Long> percentiles) {
super(percentiles.size());
putAll(percentiles);
}
}

View File

@@ -8,6 +8,7 @@ import java.util.Map;
import org.lucares.pdb.plot.api.AggregatorCollection;
import org.lucares.pdb.plot.api.Limit;
import org.lucares.pdb.plot.api.Percentiles;
public interface DataSeries {
public static final Comparator<? super DataSeries> BY_NUMBER_OF_VALUES = (a, b) -> {
@@ -35,6 +36,8 @@ public interface DataSeries {
public double getAverage();
public Percentiles getPercentiles();
public void setStyle(LineStyle style);
public LineStyle getStyle();
@@ -114,5 +117,4 @@ public interface DataSeries {
return result;
}
}

View File

@@ -1,6 +1,8 @@
package org.lucares.recommind.logs;
import org.lucares.pdb.plot.api.AggregatorCollection;
import org.lucares.pdb.plot.api.CumulativeDistributionCustomAggregator;
import org.lucares.pdb.plot.api.Percentiles;
public class FileBackedDataSeries implements DataSeries {
@@ -55,11 +57,19 @@ public class FileBackedDataSeries implements DataSeries {
@Override
public double getAverage() {
return csvSummary.getStatsAverage();
return Math.round(csvSummary.getStatsAverage() * 10.0) / 10.0;
}
@Override
public AggregatorCollection getAggregators() {
return csvSummary.getAggregators();
}
@Override
public Percentiles getPercentiles() {
return csvSummary.getAggregators()//
.getAggregator(CumulativeDistributionCustomAggregator.class)//
.map(CumulativeDistributionCustomAggregator::getPercentiles)//
.orElse(new Percentiles());
}
}

View File

@@ -2,12 +2,15 @@ package org.lucares.pdbui.domain;
import java.util.Collection;
import org.lucares.pdb.plot.api.Percentiles;
public class DataSeriesStats {
private final int values;
private final long maxValue;
private final double average;
private final String name;
private final StyleAndColor dashTypeAndColor;
private Percentiles percentiles;
public DataSeriesStats(final int values, final long maxValue, final double average) {
this.name = "<noName>";
@@ -17,13 +20,14 @@ public class DataSeriesStats {
this.average = average;
}
public DataSeriesStats(final String name, final StyleAndColor dashTypeAndColor, final int values,
final long maxValue, final double average) {
public DataSeriesStats(final String name, final StyleAndColor dashTypeAndColor, final int values, final long maxValue,
final double average, final Percentiles percentiles) {
this.name = name;
this.dashTypeAndColor = dashTypeAndColor;
this.values = values;
this.maxValue = maxValue;
this.average = average;
this.percentiles = percentiles;
}
/**
@@ -47,14 +51,18 @@ public class DataSeriesStats {
return average;
}
public Percentiles getPercentiles() {
return percentiles;
}
public String getName() {
return name;
}
@Override
public String toString() {
return "[name=" + name + ", dashTypeAndColor=" + dashTypeAndColor + ", values=" + values + ", maxValue="
+ maxValue + ", average=" + average + "]";
return "[name=" + name + ", dashTypeAndColor=" + dashTypeAndColor + ", values=" + values + ", maxValue=" + maxValue
+ ", average=" + average + "]";
}
public static double average(final Collection<DataSeriesStats> stats) {

View File

@@ -79,7 +79,7 @@ public class PlotResponseStats {
dataSerie.getStyle().getPointType());
dataSeriesStats.add(new DataSeriesStats(dataSerie.getTitle(), lineStyle, dataSerie.getValues(),
dataSerie.getMaxValue(), dataSerie.getAverage()));
dataSerie.getMaxValue(), dataSerie.getAverage(), dataSerie.getPercentiles()));
}
final double average = Math.round(DataSeriesStats.average(dataSeriesStats));