MAPPING AN ACBL DATABASE
A MAP CAN BE MORE THAN A PRETTY ILLUSTRATION:  IT IS A POWERFUL RESEARCH TOOL THAT CAN DIRECT QUESTIONS AND ANSWERS
Sandy Arlinghaus, sarhaus@umich.edu

General Overview

Look over my shoulder as I map the database obtained from the ACBL.  Rick Beye, Carol Robertson, Ed Evers, and Richard Oshlag have been extremely helpful in specific ways and Jay Baum has been helpful in general ways. 

There might be many reasons to map a database.  One reason is that all databases have errors in them.  Looking at them from different vantage points enables one to catch errors and correct them.

The database I'm working with has information about each ACBL district.  It's separated into subsets, one for each of the 25 districts.  Each subset contains zipcodes within each district and also units within each district.  That is the general idea.  Maps are made from zipcode information.  When all the zipcodes are partitioned into the 25 separate districts, the map appears as in the first image below (Figure 1).  The zipcode mapping is for the US only (what was in the database).  I've put other maps for Canada and Mexico in, as well, to present the full picture.  The Yukon was assigned to District 19 (in pink) and the Northwest Territories were assigned to District 18, in blue, because a check of the member database showed that that was where the majority of members in those regions had been assigned.


Figure 1.  ACBL land---notice the gaps (in black) in the underlying database.  The map is a picture of a database.

Figure 1 is quite revealing when considered as a picture of an ACBL database.  Here are a few items to consider, all of which involve issues of assignmet of zip code polygons to ACBL districts (from the database I'm using...):
Take a closer look using Figure 2, below.

Figure 2.  A closer look at Figure 1 to illustrate the effects noted above:  Swiss Cheese, Edge, and Crumbs.  Colors are chosen to enhance these effects; a pretty map can also be made but that is not the point...the map is being used as a research tool to ask questions, here.

How might we make the map better, and why should we care about doing so?

Figure 3.  Notice a variety of Crumb effects (in light purple, light brown, olive/gold, and dark green)...some farther-flung than others.  There are also Swiss Cheese effects (in black) and one combined Edge and Crumb effect (in dark green)


It seems likely that in a large database, one would not see these effects simply by glancing at the database (or would even know to look for them).  The map shows one quickly what to look for in the database.  Further, data consolidation is important because it can allow management to streamline the handling of information.  When computers are used appropriately to handle information, they enhance opportunities for humans to do what they do best--computers are a tool, and a powerful one at that, that serve humans (they do not replace them or reduce a need for personal interaction...they enhance those opportunities when used appropriately--a power saw in the hands of an amateur is a dangerous weapon; in the hands of a master carpenter it enhances design opportunities).

Greater detail for those interested....

The ACBL database has fields (columns) for ZIPcode and Unit.  I've added an extra one for District, for mapping purposes.  Many of the Unit entries are coded as "0" and for these entries I have assigned the zipcode areas to a district.  Thus, for example, prior to the assignment of District 12 "0" unit entries, the map of that district (in green) appears as in Figure 4.  The black holes show where there is no data.  Yet, we who live here know that then all of the Detroit metro area and most of the ACBL members of District 12 are left out.  The reasons for this involve concerns of conflicting cultural viewpoint that are hopefully now relegated to the past.  However, residents of the two units involved are free to join either one and are not assigned by ACBL, on the basis of zipcode location to either one.


Figure 4.  District 12, in green, without the adjustment of assigning zipcodes with a "0" code.

When the appropriate assignments are made, the result appears in Figure 5, below.  Correction of this particular assignment problem was easy.  There are numerous assignment problems that are not that easy (at least for me...they may be for locals in the districts involved) particularly on the west coast.  Adjustment of the database to reassign "0" zipcodes is a substantial task and one that a committee might ask be done in the interest of accuracy of database which in turn translates to efficiency of service to affected units.  I've given it my best shot and the results show in Figures 1 and 2 above. 


Figure 5.  Assignment of "0" polygons presents a consolidated view, at the district level.

The map in Figure 5 might suggest that the assignment made solved all problems.  It did, but only at the district level.  When one asks the computer to make the map at the Unit level, there is still the same difficulty as noted above because zipcode polygons are not assigned to individual units.  The map in Figure 6 illustrates this problem.  The darkest brown area is really two units displayed as one (notice also some Crumb effects).


Figure 6.  District 12 by units...but, there are actually five units in District 12, not the four displayed here!  Click here to go to a Clickable map of the district--once there, click on zip code polygons to bring up associated database entries.

The solutions are that one must either assign zipcode polygons by unit or merge the involved units.  Those solutions, however, may not be palatable to locals even if they are to others from a central management perspective or from an efficiency of information flow standpoint.

In brief, maps can guide the decisions we make and the decisions we make can guide the maps we create.

Let me know the decisions you make about mapping, therefore!

I will be updating this page as I produce more maps.  In the near future, I may map some population (various data sets can be linked to the map to analyze zipcode areas by the variables within the linked data sets) and I may make a better looking map, assigning black holes to districts (but leaving the crumbs alone) and adjusting the color pattern to a less abrupt one (more visually pleasing but less revealing in terms of errors).  I may also put zipcode generated maps of districts on Bill's website, in addition to the county generated maps already there, and I may make unit maps for each district (but there will be many holes coming from the 0 zipcode areas...still, it might be interesting).

There are other maps I've made on the following links:

http://www-personal.umich.edu/~copyrght/wca/ 

go to the "Atlas" button (has older district boundaries based on counties rather than newer ones based on zipcodes).
on Bill's Director website--when you look at these, think about charts, such as the pie charts, you might like on maps or any other analysis tools you might like...

and

http://www-personal.umich.edu/~copyrght/wca/AileenO/ClickClubMap2.html

a link to a clickable map made for Aileen Osofsky, also as a link on Bill's website above...again using older district boundaries but showing how a map can interact with a database, online, and with the user interaction controlling what appears on the screen.  When you look at this one, think about what other clickable maps you might like.