I sought out to take a look at the distribution of banking locations across and around New York State. I had some preconceptions that bank location would follow population density around the state, but what I was more interested was the cold spots, or area that lack access to a bank and need to travel for financial services. Initially i thought this would be simple as I came across a shapefile with all FDIC registered bank locations in New York State.

However, once I imported the file into ArcGIS Map the trouble began. As the file was already a shapefile I was unable to filter by attribute (specifically by county) which lead me figuring out how to reformat shapefiles. Eventually I figured out that I needed to export the shapefile as a text file and then filter the data in excel. This seemed like a time consuming and frustrating process but I was unable to find an alternative method within ArcGIS aside from copying the data from the target column and pasting it in another newly created column that was filterable within ArcGIS.

Once I had all of the data in a csv the remainder of the lab flowed smoothly. I split the data into two categories, the NYC area and Upstate New York. The data was split at this level because when I ran an optimized hot spot analysis in ArcGIS (Getis-Ord Gi*) and the entire NYC region was a hot spot. Once the data was separated you were able to see the cold spots in NYC and across the state.

Within NYC, the analysis turned up a number of hot spots, with all being either in Manhattan or directly adjacent to Manhattan. Additionally parts of the Bronx, Brooklyn and Queens all have a few cold spots. More work could be done here to compare the banking cold spots against ACS data for income or home-ownership and determine if there are any credit deserts present in NYC.

State level analysis was not as productive, there were three main hotspots identified by the analysis and unfortunately all three of the hotspots are in population centers, Albany, Syracuse and Buffalo. One thing lacking in this analysis is the lack of data distributed across the state. As there simply is no data for vast portions of the state the analysis does not have any values to input into the algorithm and the results come back as not significant.