Applies to: @RISK, all versions

Why is the mode far from the tallest histogram bar in @RISK?

 

Answer: 

The mode of continuous data is computed differently from a histogram bar. For discrete data, the mode is simply the most frequent value. For continuous simulation results, identical points are rare, so @RISK uses a binning method:

1. Divide the simulated data into k = 100 bins (or k = n/3 if fewer than 300 points).

2. Each bin has equal numbers of points; bin widths vary.

3. The narrowest bin (points most clustered) is chosen.

4. The mean of points in that bin is reported as the mode.

Histogram bins, in contrast, all have the same width and may contain different numbers of points. This difference in binning explains why the mode can be far from the tallest bar.

@RISK uses the same method when displaying approximate modes for fitted distributions, but only for comparison purposes—not for fitting.

 

Attachments:
ModeOfContinuous.xls