hudson.graph.jfreechart
Class Utils

java.lang.Object
  hudson.graph.jfreechart.Utils

public class Utils
extends Object

Chart generation utility code around JFreeChart.

Author:

Kohsuke Kawaguchi

See Also:

StackedAreaRenderer2, DataSetBuilder, ShiftedCategoryAxis

Field Summary

static double

CHEBYSHEV_N

Constructor Summary

Utils()

Method Summary

static void

adjustChebyshev(org.jfree.data.category.CategoryDataset dataset, org.jfree.chart.axis.NumberAxis yAxis) Adjusts the Y-axis so that abnormally large value won't spoil the whole chart by making everything look virtually 0.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

CHEBYSHEV_N

public static double CHEBYSHEV_N

Constructor Detail

Utils

public Utils()

Method Detail

adjustChebyshev

public static void adjustChebyshev(org.jfree.data.category.CategoryDataset dataset,
                                   org.jfree.chart.axis.NumberAxis yAxis)

Adjusts the Y-axis so that abnormally large value won't spoil the whole chart by making everything look virtually 0.

The algorithm is based on Chebyshev's inequality, which states that given any number sequence, nore more than 1/(N^2) values are more than N x stddev away from the average.

So the algorithm is to set Y-axis range so that we can see all data points that are within N x stddev of the average. Most of the time, Cebyshev's inequality is very conservative, so it shouldn't do much harm.

When the algorithm does kick in, however, we can kick out at most 1 in N^2 data points. (So for example if N=3 then we can "fix" the graph as long as we only have less than 1/(3*3)=11.111...% bad data.

Also see issue #1246.