Reputation: 6783
I'm pretty much new to JFreeChart and am facing a performance issue while trying to render a huge dataset. The dataset in my scenario comes from a csv file which has well over 46700 rows (lines) and around 2666900+ data points to plot.
My current code for plotting is as follows:
public void doPlot(String plotTitle, HashMap<Integer, ArrayList<PlotCheckBox>> plotDataList, boolean dotOnly) {
plotDialog.dispose();
DefaultXYDataset xyDataSet;
try {
if (plotDataList.isEmpty()) {
return;
}
xyDataSet = new DefaultXYDataset();
/**
* Populate Data Set
*/
boolean[] optionList;
int countPlotPoints = 0;
for (Integer tabIndex : plotDataList.keySet()) {
ArrayList<PlotCheckBox> checkBoxList = plotDataList.get(tabIndex);
try {
if (checkBoxList == null) {
continue;
} else if (checkBoxList.isEmpty()) {
continue;
}
optionList = new boolean[checkBoxList.size()];
for (int index = 0; index < checkBoxList.size(); index++) {
optionList[index] = checkBoxList.get(index).isSelectedForPlot();
}
/**
* The getTabXYData() method builds up the xyDataSet by loading values from a given csv file and the
* attributes chosen by the user to plot
*/
countPlotPoints += getTabXYData(xyDataSet, tabIndex, optionList, jTabbedPane_results.getTitleAt(tabIndex));
} finally {
checkBoxList = null;
optionList = null;
}
}
System.out.println("Plot Points In This Graph: "+countPlotPoints);
if (countPlotPoints == 0) {
print("No options selected.\n");
JOptionPane.showMessageDialog(this, "No Plot Points Were Selected!", "Warning", JOptionPane.WARNING_MESSAGE);
return;
}
JFreeChart chart = ChartFactory.createXYLineChart(plotTitle, "time(ms)", "---", xyDataSet, PlotOrientation.VERTICAL, true, true, false);
XYPlot plot = (XYPlot) chart.getPlot();
XYLineAndShapeRenderer renderer = new XYLineAndShapeRenderer();
for (int i = 0; i < countPlotPoints; i++) {
if (dotOnly) {
renderer.setSeriesLinesVisible(i, false);
} else {
renderer.setSeriesLinesVisible(i, true);
}
renderer.setSeriesShapesVisible(i, true);
}
plot.setRenderer(renderer);
ChartPanel chartpanel = new ChartPanel(chart);
chartpanel.setDefaultDirectoryForSaveAs(new File(lastAnalyzedPath));
JFrame frame = new JFrame();
frame.setTitle(plotTitle);
frame.add(new JScrollPane(chartpanel));
frame.pack();
frame.setVisible(true);
frame.addWindowListener(new java.awt.event.WindowAdapter() {
@Override
public void windowClosing(java.awt.event.WindowEvent evt) {
try {
System.out.println(":: Clearning Memory ::");
System.out.println("\tFree Memory (Before cleanup): "+Runtime.getRuntime().freeMemory());
Component component = getComponent(0);
if(component instanceof ChartPanel){
JFreeChart chart = ((ChartPanel) component).getChart();
XYPlot plot = (XYPlot) chart.getPlot();
plot = null;
chart = null;
component = null;
}
} finally {
System.runFinalization();
System.gc();
System.out.println("\tFree Memory (Post cleanup): "+Runtime.getRuntime().freeMemory());
}
}
});
} finally {
xyDataSet = null;
System.runFinalization();
System.gc();
}
}
Because of this huge dataset, the plot takes a lot of time to load and trying to change the size of the plot window throws an OutOfMemoryError and the application crashes.
What I would like to know are suggestions for improving performance. Here's what I've thought of (any comments/suggestions/feedback on this would really be appreciated):
Restricting users to a specific range to plot. Problem is I don't know how much of data can JFreeChart handle. Any suggestion on this? I would preferably won't want to use a magic number or just a trial and error method.
Use a scrollable XYDataSet. I'm pretty much new to this and don't have much idea about the implementation. Any sample code and a comment on efficacy of using this technique would be highly appreciated.
I'm open to exploring new ideas. Please do let me know what you think about this issue. Many a thanks in advance!
Upvotes: 1
Views: 1708
Reputation: 205805
For ~106 data points, FastScatterPlot
is a good choice. For larger numbers, you'll have to test. For speed, it uses an internal render()
method, instead of a plug-in renderer. Also consider using a SwingWorker
to update the chart incrementally as data is read.
Upvotes: 1