scatter-plot of 300k points is very slow about incanter

[nathanmesser]

Have you got some sample code that tries to do this and is very slow?

[nathanmesser]

I've found that I can get the speed on creating a scatter plot for 300k items down from around 20s on my machine, to almost instant, by switching from an XYSeriesCollection to a DefaultXYDataset. However I need to do more digging to see what the downsides of this would be.

[alexott]

@nathanmesser do you have any code snippets that allows us to use DefaultXYDataset?

[MawiraIke]

I've found that I can get the speed on creating a scatter plot for 300k items down from around 20s on my machine, to almost instant, by switching from an XYSeriesCollection to a DefaultXYDataset. However I need to do more digging to see what the downsides of this would be.

Has the scatter-plot method been optimised already? It seems like it creates a plot of 300k items in about 577ms on the first run on my side. The second and subsequent runs take about 200ms each. Might I be measuring a different time? I am using

(time (view (scatter-plot data)))

I am about to test if DefaultXYDataset is faster, I have time i can throw at it if there is no progress yet. I see that scatter-plot still implements XYSeriesCollection.

[MawiraIke]

Using a DefaultXYDataset takes fairly the same amount of time on my side as the XYSeriesCollection but while i was doing more research, i stumbled across the AbstractXYDataset and FastScatterPlot which might both be faster. I don't think it's necessary to implement them as already the implemented solution is 'fairly' fast.