Clickstream

BI Tools for Gerrymandering

This excellent article in the New Yorker describes the current redistricting process and the gerrymandering that's taken root. What I found intriguing is the use of Maptitude software from Caliper. By feeding as much census and election data as possible into the software it is possible to point-and-click your way to a safe congressional seat in the next election.

With a few clicks, Persily changed the map from one that showed party registration in each census block to one that revealed voting results in each block. The colors ranged from dark red, for heavily Democratic votes, to dark blue, for strongly Republican. He showed voting results in about two dozen races, from President to governor and from congressman to local offices. "The whole process has got much more sophisticated" Persily went on. "Party-registration data are not the only kind of data you want to use. You want to use real election results. That's a big change from ten years ago. We have become very good at predicting how people are going to vote. People's partisanship is at a thirty-year high. If I know you voted for Gore, I am better able to predict that you are going to vote for any given Democrat in a future election."

The use of Maptitude reminds me of hospital third-party billing software I worked on in the mid-eighties. The primary, secondary and tertiary diagnoses determine the allowable rates for insurance and medicare billing. In general, one combination provides a better total rate than the others. It was illegal for a computer system to determine the ordering, but the person doing the data entry could. So they asked us to write software that would provide every combination of diagnostic combinations and display them with the billing amounts. The user then selected the one that had the highest rate. While not illegal it definitely subverted the intent of the laws at that time.

This redistricting software does much the same thing to the political redistricting process. There was never an intent to let politicians in power redistrict to make it impossible for them to lose elections. Now that computers and good BI software make it as simple to geryymander as clicking a mouse, we need laws that explicitly forbid this practice. The problem is that two thirds of politicians in office hold safe seats because of the subverted process, so it's unlikely that any reform will take place.

I'm also embarrassed to note that my own alma mater (CMU) played a part in the Pennsylvania redistricting that's now in court.

Posted by Mark Friday, December 05, 2003 8:16:00 PM | permalink | (0) comments

MySQL Gotchas

I've only worked with MySQL a little bit, so I'm not yet comfortable using it for data warehousing. I'll be researching Open Source databases in a lot more detail this spring in advance of a few conferences, with the goal of pumping some large volumes of data into them to see how they handle the usual range of business intelligence tasks. Expect a spate of magazine articles come mid-summer on the results, assuming my hardware is set up in time.

While catching up on some reading I came across this useful page about some of the more annoying features of MySQL. The author doesn't treat these as bugs since they are mostly documented and designed-in behavior. For example:

If no DEFAULT value is specified for a column, MySQL automatically assigns one, as follows. If the column may take NULL as a value, the default value is NULL. If the column is declared as NOT NULL, the default value depends on the column type: (...)

I did not realize that MySQL's treatment of NULL values was as strange as the examples he walks through. Worth a read if you are starting to do any development work with MySQL (not just DW experimentation) and you are already familiar with commercial and standards-compliant databases.

Posted by Mark Wednesday, December 03, 2003 11:14:00 PM | permalink | (0) comments

Excel for Visually Immersive BI

No BI tool is complete until it can handle sophisticated graphical applications. I've only recently learned that Excel makes the grade, thanks to this nice demo of a sophisticated GUI(291K).
Learning how to do this makes spreadsheets much more fun, sorta like when Visicalc was first released.
[Note: there has been some trouble downloading the file from it's original source so it's mirrored here just in case]

Posted by Mark 10:40:00 PM | permalink |

The Latest Outsourcing Trend: Primate Programming

The logical choice to lower outsourcing costs is to find a less expensive labor pool. Primate Programming has the ultimate solution to the problem.

Here by the San Diego zoo, experiments last month with baboons have proved that higher primates can perform software testing, traverse complex menus, and code simple XML schemas. The finding have implications for the entire software industry, with some scientists predicting routine programming such as maintenance and report writing will be performed by teams of primates within 10 years. McAuliffe’s work builds on research conducted in 2003 at several research universities. This university research supports the view that higher primates can learn language and perform complex cognitive tasks.

The next time someone asks you about outsourcing your data warehouse or business intelligence project, you''ll know where to go.

Posted by Mark Friday, November 14, 2003 10:03:00 PM | permalink | (0) comments

Speaking at the Data Warehouse Institute Conference in May

I'll be doing a half-day course on ETL tool selection at the TDWI conference in Boston in May. As I develop materials, I'll be posting some of them here. In the meantime, if you are interested in evaluating data extraction tools for data warehousing, I refer you to the ETL product checklist I wrote on the topic> last year.

Posted by Mark Tuesday, November 11, 2003 11:00:00 PM | permalink | (0) comments