Monthly Archives: September 2015

SPERT-Beta Development Release B

This is version 0.1 Build 11.  The change between Build 10 and 11 included a major change.  While testing, I learned that I wrongly set the order of the alpha and beta shape parameters between P4:Z15, so the implied shape of the curve was skewed on the opposite side of where it should have been.  The formula changes, then, correctly set the alpha and beta shape parameters so when a 3-point estimate implies a left-skewed curve, the shape parameters create an implied, left-skewed curve (and vice versa for a right-skewed curve).

There were a few minor changes, too:  I vertically centered all cells and added a minor edit to one column to avoid a #VALUE error if the wrong order of input values is used.

(Visit the Download page to download the latest version of Statistical PERT – Beta Edition).

SPERT-Beta Development Release A

If you’d like to get a sneak peek at the next edition of Statistical PERT which uses the beta distribution, you’ve come to the right spot!  This still-under-development release of the SPERT-Beta Excel workbook will only be available through this blog, not through the download page of the Statistical PERT website (which currently offers SPERT downloads using Excel functions for the normal distribution).

This SPERT-Beta workbook works just like all the other SPERT workbooks and templates.  Just add a 3-point estimate and make a subjective opinion about how likely the most likely outcome really is.  That’s it!  The SPERT-Beta formulas correlate your 3-point estimate to a matching set of beta distribution shape parameters which then can be used in Excel’s BETA.DIST and BETA.INV functions.

There’s a lot I could say about how this template is constructed, and that will come out in time.  Right now, if you download and use this template, feel free to share with me any issues, questions or problems you have.  Keep in mind that this is a work-in-progress, and visit here often to obtain the latest SPERT-Beta downloads.

(Visit the Download page to download the latest version of Statistical PERT – Beta Edition).

Statistical PERT – Beta!

This blog will provide information and download information about a new edition of Statistical PERT using the beta probability distribution!

I built the original Statistical PERT version in 2014 to easily leverage the built-in statistical functions in Microsoft Excel.  Among those statistical functions, the functions for the normal probability distribution (the so-called “bell-shaped curve”) — which are NORM.DIST and NORM.INV — are very easy to use, provided you know the standard deviation for a bell-shaped uncertainty.  Statistical PERT provides a rationale way for obtaining a standard deviation that corresponds to the bell-shaped uncertainty and the estimator’s subjective opinion about how likely the most likely outcome really is.

Using the normal distribution with Statistical PERT is very easy to do.  But the normal distribution isn’t appropriate for uncertainties which are more than just mildly skewed.

The beta distribution is the actual distribution upon which the original PERT was based.  And Microsoft Excel has built-in functions for the beta distribution, just like it has for the normal distribution (BETA.DIST and BETA.INV).  However, the beta distribution functions require several arguments, but the most difficult two arguments are two shape parameters, alpha and beta, that dictate the probabilistic shape of the resulting beta distribution.  Using these shape parameters, one can create a beta distribution that looks identical to a normal, bell-shaped distribution, or you can create right triangles, or shapes that show very strong skewing either to the left or to the right.  In short, the beta distribution is very flexible, but it’s harder to use.

My goal is to expand Statistical PERT to offer both an edition that’s easy to use, understand, and manipulate using the normal distribution, and another edition which is still just as easy to use and understand, but is more accurate and flexible than the version using the normal probability.  To do that, though, it will not be as easy to calibrate the SPERT templates which use the beta distribution.  It can still be calibrated and customized, but just not as easily.

Very soon, I’ll begin making downloads available on the new edition of Statistical PERT which uses the beta distribution.  These downloads won’t even be “beta” software, rather, they’ll be work-in-progress, “alpha” software which haven’t been tested rigorously and which will change and evolve a lot before I make them available for downloading on the Statistical PERT regular download page.

I’m really excited about expanding Statistical PERT in 2016 to a wide variety of project managers and anyone else who needs to easily make probabilistic estimates.  And I’m really excited about offering an edition of Statistical PERT which leverages the beta distribution inside Microsoft Excel!