City level analysis (shared zip codes between cities)

Hello,

Thank you for reaching out on the community forum! I would be glad to offer some insight in to this.

Unfortunately, there isn't an ideal solution when deciding which city should be assigned the zip code whenever they overlap since it's rather unusual to do a study of such closely linked regions but I understand this can be necessary at times. We do have some suggestions for your best options (with accompanying drawbacks): 

1) You can leave the zip code out of the model entirely to avoid the issue altogether. This will most likely reduce your impact if the industries you are studying are in that zip code. You may also run in to more serious issues if that industry is ONLY in that zip code in the specific city you are studying.

2) You can contribute the zip code to the city which contains the majority share of the population. The drawback here is that your impact may be geographically closer to region B but you'll be assigning the zip code to region A. For instance, you may have a zip code in City A and City B. Perhaps 70% of the population lives in City A so you assign that zip code to City A. But then your study is on an industry only located in City B's part of the zip code. You'd be incorrectly assigning your impacts to the wrong city. 

3) You can assign the zip code to the city most economically-correlated with it. For instance, if a shared zip code has a lot of industries that support your primary impact, you might find better results by pairing it accordingly. This would require familiarization with the study area in question. 

4) You can build the model both ways (the shared zip code in both City A and City B), view the impacts, then attribute the zip code to the city with the largest benefit. You may find that the difference is marginal, depending on the geographic distribution of the supporting industries.

I hope this helps and wish there was a more concise solution but unfortunately there isn't a single "best" option. 

Thank you!