Re: Strange Result from Histogram in V6

*To*: mathgroup at smc.vnet.net*Subject*: [mg83306] Re: Strange Result from Histogram in V6*From*: Norbert Marxer <marxer at mec.li>*Date*: Sat, 17 Nov 2007 05:20:31 -0500 (EST)*References*: <fhjs1s$4qn$1@smc.vnet.net>

On 16 Nov., 11:39, Scott <sguth... at gmail.com> wrote: > The command: > > Histogram[{1/4, 1/3, 1/2, 1/4, 1, 1/4, 1/2, 1/3, 1/4}, > HistogramCategories -> {0, 1/4, 1/3, 1/2, 1, 2}] > > seems to find 48 values between 0 and around .2 and 12 values between > around .2 and .5 in Version 6. > > Very strange. > > Any insight? > > Cheers, Scott Hello Note that the default setting for HistogramScale is Automatic, which - for unequally sized intervals - behaves in the same way as HistogramScale->True. As you can read in the documentation center: HistogramScale -> False does not scale the bar heights and therefore gives the number of measurements for each interval. HistogramScale -> True scales the bar heights by the interval width. HistogramScale -> 1 gives a probability density plot. I would also use HistogramRange->{0,2}. Then you can see better that the first interval [0,1/4] is empty. Without scaling: Histogram[{1/4, 1/3, 1/2, 1/4, 1, 1/4, 1/2, 1/3, 1/4}, HistogramCategories -> {0, 1/4, 1/3, 1/2, 1, 2}, HistogramRange -> {0, 2}, HistogramScale -> False] With scaling: Histogram[{1/4, 1/3, 1/2, 1/4, 1, 1/4, 1/2, 1/3, 1/4}, HistogramCategories -> {0, 1/4, 1/3, 1/2, 1, 2}, HistogramRange -> {0, 2}, HistogramScale -> True] You get the number 48, because you have 4 measurements in the interval [1/4,1/3[ with a width of 1/12. Best Regards Norbert Marxer