Histogram countsFracCurve should not be normalized by default

I’m no statistics guy but is this correct behaviour?

I would have expected the distribution curve to be scaled accordingly to the histogram, but it seems to only work if the histogram range of values is 100 exactly.

var plt = new ScottPlot.Plot(600, 400);

Random rand = new Random(0);
double[] values = DataGen.RandomNormal(rand, pointCount: 10000, mean: 50, stdDev: 5);
hist = new ScottPlot.Statistics.Histogram(values);

double barWidth = hist.binSize * 1.2; // slightly over-side to reduce anti-alias rendering artifacts

plt.PlotBar(hist.bins, hist.countsFrac, barWidth: barWidth, outlineWidth: 0);
plt.PlotScatter(hist.bins, hist.countsFracCurve, markerSize: 0, lineWidth: 2, color: Color.Black);
plt.Title("Normal Random Data");
plt.YLabel("Frequency (fraction)");
plt.XLabel("Value (units)");
plt.Axis(null, null, 0, null);
plt.Grid(lineStyle: LineStyle.Dot);

plt.SaveFig("Advanced_Statistics_Histogram.png");

[LB767]

Awesome! Looking very much forward to the next major release (multiple Y axes 😄)

[swharden]

Just fixed this in 4.0.44 (on NuGet now) and it works as expected

var hist = new ScottPlot.Statistics.Histogram(values);
plt.PlotBar(hist.bins, hist.countsFrac, barWidth: hist.binSize * 1.2, outlineWidth: 0);
plt.PlotScatter(hist.bins, hist.countsFracCurve, markerSize: 0, lineWidth: 2, color: Color.Black);

Thanks again for noting this issue! Scott