Why did Power Map plot Lake County, Colorado, in Indiana?
Why did Power Map plot Lake County, Colorado, in Indiana, adjacent to Lake Michigan? In the Power Map below, the extra tall, beige, bar over Indiana really belongs in Colorado. The tall bars in Alabama and Maryland also are out of place. What’s going on here?
The data bars appear in the “wrong” spots because that’s where you — well, me in this case — placed them. To geocode correctly, Power Map needs unambiguous location data. Microsoft says so right in the documentation. In the US, there are 12 counties named “Lake.” That leaves PowerMap with a lot of room for interpretation. So it did just that.
￼￼To avoid the ambiguity, there’s a right way and wrong way. The Geography dialog shown below seems ok, but really is not. By geocoding using both the State and County values – pulled from distinct columns in my dataset – you might guess that Power Map would differentiate all the Lake counties. It doesn’t. Plus, it gives no indication of the problem. I discovered it when I overlaid a chart element onto the map, one that in this odd case, happened to expose extreme outliers.
The best way to avoid ambiguity is to feed Power Map truly unique geographic values *in a single column*. However, while this rule seems straight forward enough, it does leave some room for interpretation. As it turns out, that’s OK.
The following list demonstrates four variations on how to structure the location data, using Lake County, Colorado, as an example. Each one representing a valid and unique county name in the US, from the point of view of Power Map.
1. Lake Colorado
2. Lake CO
3. CO Lake
4. Colorado Lake
Case #1 is the best structure, but is only slightly better than #2. Both structures result in 100 percent of County names in the US being geocoded. The difference is that #1 yields a slightly higher match rate at the “high confidence” level: 99 percent vs 98 percent. While #3 and #4 mostly work, you end up with a few dozen counties that are not unrecognized.
Notice that I did not include the qualifier “County,” such as “Lake County, CO.” It’s unnecessary, and can in fact lower the geocoding success rate. Also notice the lack of a comma delimiter, such as “Lake, CO.” It too is unnecessary and Power Map appears ignore it anyway. In the snippet shown below, from my data model, the column “County Map ID” demonstrates the preferred structure.
If you dataset covers both the US and Puerto Rico, your might be tempted to add the Country identifier into the label, such as “Lake County, Colorado, USA.” Don’t though because you’ll actually impede the geocoding process. Instead,you should include the Country identifier as a second geography type, like in the Geography dialog shown below.
Sometimes the simplicity of the tools gets the better of you. Always be sure to study the data for accuracy and don’t just rely on the lack of error messages to conclude that your project is correct.