Cumulative Distribution Function (CDF)

Wed, 02/13/2013 - 12:17 pm

The code below will estimate the cumulative distribution function, P(X≥x) from a wave of observations. There is an option to plot the complementary cumulative distribution function, P(X≤x), as well. Note that an alternative way to calculate this would be to create an ordinary histogram and integrate. The version presented here is without binning. The two advantages are that (1) you don't have to worry about optimal bin size and (2) you do not lose any information.

The attached example program shows typical results.


Function MakeCDF(w,sortflag)				// creates empirical cumulative distribution function
	wave w; variable sortflag
	variable n=numpnts(w)
	string wroot=NameofWave(w),cdfname=wroot+"_cdf"
	duplicate /o w $cdfname
	wave cdfwave=$cdfname
	
	if (sortflag == 1)
		Sort /R cdfwave cdfwave				// largest values first (for complementary CDF)	
	else	
		Sort cdfwave cdfwave				// smallest values first (for CDF);  default case
	endif

	SetScale x 0,1,"", cdfwave				// fraction of x values
End

MakeCDF.pxp (259.29 KB) MakeCDF_0.pxp (259.29 KB)

Forum

Support

Gallery

Cumulative Distribution Function (CDF)

Igor Pro 10

Igor XOP Toolkit

Igor NIDAQ Tools MX