constructed with Crystal Ball's Custom Distribution
The distribution is useful in a non–parametric technique for replicating the distribution shape of a large set of data. The technique is simply to collate the data into i bands that each have a minimum and maximum you specify, calculate the number of data values that fall into each band, and then use this information to define the distribution. It has the disadvantage of 'squaring off' into the histogram shape, but with a lot of data and small bands the technique is a transparent and practical way of fitting a distribution to data.
Crystal Ball automatically normalizes the distribution to have an area of one.