Imperfect Information: The Decision Analysis Blog

The 50th birthday of Decision Analysis led to an interview (podcast and transcript here) of Ron Howard in the HBR IdeaCast series. Judging from social media it's been attracting some attention in the DA community, and one hopes more broadly. This is the sort of thing I'd like to see more of, DA mavens interacting with members of the media who are business oriented but not part of the DA or even OR/MS choir.

Every DPL model is a combination of an Influence Diagram and a Decision Tree but I admit that we tend to play up the decision tree side of our software, as it is the analytic backbone of the DPL software. Furthermore, as a marketer, I'm also vested in search term prominence. See how the two terms compare in Google Trends above.

Q. Should your portfolio system be spreadsheet-based and open or database-powered and robust?
A. Yes!

In an earlier post I introduced DPMX, and said a few words about the importance of data management and the value of transparent modelling environments like Excel and DPL. Here I'm returning to the topic to drill down on the subject of spreadsheets and databases, which are close friends in our milieu.

We're back from the INFORMS 2014 Annual Meeting, and it was a doozy. If you were not able to attend, we've taken the sting out of that by giving you the high- and low-lights from the event.

Highlights:

1. Our well attended official DPMX software demo, and the on-demand demos requested in the booth. We're jazzed about DPMX and love talking about it.

Decision Analysis has reached a certain age. As with most ideas, decision analysis came about over a period of time, but the official methodological dating committee has deemed Ron Howard's 1964 paper, "Decision Analysis: Applied Decision Theory" to be the day the stork arrived. To celebrate, DAS and SDP organized a pre-informs session featuring talks, drinks and dinner, held the day before the start of the INFORMS annual meeting (call it INFORMS day zero).

In recent months we at Syncopation have been hard at work on a set of tools we call DPMX, short for DPL Portfolio Management eXtentions. DPMX is a key piece of the puzzle for organizations that want to do portfolio analysis/prioritization based on decision analysis principles and don't want to get lost in a labrynth of email, spreadsheets and PowerPoint decks.

The DPL maintenance update 8.00.07 is now available for download. This update includes several enhancements and bug fixes, including these:

BUG: Policy tree outlining shows extra rows (Excel side interface) (#1792)
BUG: Confusing behaviour creating nodes with invalid suggested cell names (Excel side interface) (#1752)
ENH: Option to turn off Excel screen updating during linked run (#1797)
ENH: Alternate keyboard shortcuts for cut/copy/paste in get/pays and other grids (#693)
BUG: Internal error 94 updating reports with certain runtime errors (Portfolio version) (#1761)

The Wall Street Journal reported recently about a paper that purports to show people are better able to tolerate the "pain" of a difficult decision if they've taken acetaminophen, the common pain killer.

October is a lively time of year at Syncopation Software. As the leaves start dropping at our home base in New England, business tends to pick up. In addition to our code monkeys (i.e., everyone but myself) diligently plugging away on DPL, our preparations for exhibiting at the INFORMS Annual Meeting are in full swing, while a multitude of inquiries from the many students prompted by their professors to download our free DPL trial license are also being dealt with.

The DPL maintenance update 8.00.06 is now available for download. This update includes several enhancements and bug fixes, including these:

ENH: Replay endpoints enabled with a halt() function (#49)
BUG: Error creating nodes from selection two or more times when workbook is not saved (Excel side interface) (#1750)
BUG: DPL may give memory error loading large policy trees from saved workspace (#1774)
BUG: Application error in automatic update on some systems (#1775)
ENH: Support for @product function added (#1782)

If you're a member of the Decision Analysis choir, you can stop reading now. Otherwise, have you ever taken or given an almanac exercise? If not, you should! Here's what it is...

The almanac exercise is a probability assessment game well known within some DA circles, but as I keep reminding my choirboy friends, not everybody comes from ADA/SDG/DFI/EES/UT/etc.

Personnel decisions are real decisions, but rarely (if ever?) do decision analysts talk about them in the terms of the trade. Perhaps we should. To get started, I'll offer my take on the DA of HR, so to speak.

A performance review seeks to determine the performance of an employee -- how well is s/he doing the job. (Let's not get into scales, weights, etc, but assume there is a metric.) Ostensibly, that measurement is supposed to answer/inform three questions/decisions:

1. Should this employee be fired?

Get your party hats out folks. Fifty years ago our DA forefathers, Howard Raiffa and Ron Howard, independently established the field of decision analysis as it is practiced today in thousands of companies and universities across the globe. For those of you who aren't in the know, Decision Analysis, put simply, is a formal method for making decisions.

The DPL maintenance update 8.00.05 is now available for download. This update includes several enhancements and bug fixes, including these:

ENH: New/changed API functions DBDeleteData, UpdateReport, SetProjectIDs (#1726)
CHG: Prevent installation of potentially conflicting DPL versions (#1748)
ENH: Search node collection when selected (#1672)
BUG: Internal error 94 updating portfolio data reports with certain lookup errors (#1761)
CHG: Default precision for reduction increased (#1749)

(Or, Is this Decision Tree too Chunky?)

When it comes to modelling uncertainty, it seems there are two kinds of people: discrete decision tree people and continuous Monte Carlo simulation people. That's OK, to a point. Some analyses can be approached with either method, and it is just a matter of taste.

The DPL maintenance update 8.00.04 is now available for download. This update includes various bug fixes as well as some enhancements. The highlights are listed below.

All DPL users are strongly encouraged to stay current with the latest updates.

(Or, how about just a little bit of process to go with that gorgeous decision tree?)

Personally, I've always loved a beautiful model, whether it's based on decision trees, Monte Carlo, dynamic programming, or any other quantitative method. I've made a few beautiful models myself, and have admired many more created by colleagues and DPL users. However, fairly early on in my career I began to notice that the correlation between the beauty of a model and the success of a project (as measured on the armchair decision quality index) was quite small.

It's surprising to me how often I encounter "decision models" that don't actually include any decisions. I'll answer the question suggested by the title straight away: I think it's a mistake not to include an explicit decision, for reasons that will follow.

Nobody ever wants to admit that their model is a worry analysis, a pointless forecast, an anxiety management device, etc -- they will always say there is a decision. The decision is implicit, of course, can't you see it?

As the Marketing Coordinator (i.e., not programmer, analyst, consultant, etc.) at Syncopation Software, I'm the first to admit that I'm an amateur when it comes to business analytics and operations research. I've never taken a course relating to anything to do with business, advanced mathematics, decision science, operations research, etc. I've got a B.A.