(For the README file, code, and csv file please click here.)

Executive Summary

In the spirit of celebrating the long-awaited opening of legal storefronts in NYC for adult-recreational-use cannabis beginning in December 2022, I thought up the following scenario to create a map for a geoprocessing exercise using ArcGIS Pro:

You are a retail cannabis entrepreneur who’s been awarded a license by NY State and the NYC government, and your time has come to choose a retail location. You live in Queens and, therefore, would like your storefront to be located there. 

   However, your storefront cannot be located just anywhere. According to the NY State Office of Cannabis Management’s “Guidance for Adult-Use Retail Dispensaries” document, under section 23 (“Location of Licensed Premises”), assuming that the potential location of a licensed cannabis storefront complies with local zoning ordinances, it cannot be located:

1. within 500 feet of a school. (I downloaded this dataset of public school locations from NYC Open Data.)

2. within 1000 feet of another for-recreational-use or for-medical-use cannabis storefront. (Good Grades in Jamaica is the first - and so far the only - recreational-use storefront in Queens, having opened on March 30th, 2023, and my dataset of the three medical-use dispensaries that exist in Queens were compiled from www.silive.com.)

3. on the same street or avenue and within 200 feet of a building occupied exclusively as a house of worship. (I will not be applying this rule for this project, but wanted to point it out in case of future research.)

Tools Used

• Programming languages: Python, R

• Google Maps API (to geocode addresses to longitude, latitude)

• ArcGIS Pro (to perform geoprocessing and create map)

Method

1) Web scraping PropertyShark.com

First, using Python, I scraped propertyshark.com after filtering the results to show available locations only in “Queens, NY” and by “Retail” space. There were 179 results, with just 16 results appearing on each page. After inspecting the page's html code, I scraped all the "h2" divs with the class "building-name" and had the results compiled into a csv file with the column header "Address".

2) Cleaning data and geocoding addresses

Then, I went through the list and deleted a few rows where there was only ambiguous building info and no address (such as “MTA Master Leasing Opportunity”). Then, I geocoded all the addresses in R via the Google Maps API. There were 54 addresses that resulted in “NA”s, which is quite a few, but for the sake of time, I decided to remove these rows instead of trying to figure out what happened this time. In the end, I had 150 locations, but once I pulled the data into ArcGIS Pro, I noticed that seven locations were not in Queens, so I removed those and was left with 143 locations.

3) Creating buffer radius zones in ArcGIS Pro

Next, I used the Buffer tool in ArcGIS to create 500-ft buffers around existing school locations and 1000-ft buffers around the existing cannabis store locations. Then, I used the Merge tool to merge the two layers of buffers together and dissolve them to form one combined buffer. Then, I used the Dissolve tool to create a single buffer layer to be used as the Clip features to extract retail site prospects. Then, I used the Clip tool and selected my current possible retail sites as the input layer and as the Clip features selected the newly created Public Schools and Existing Cannabis Stores Dissolve layer.

   After opening up the attribute table for the newly created Clip layer, we see 23 retail locations that fall within the buffer zones. This means that these are too close to the schools and existing cannabis facilities and, therefore, they cannot be potential storefront sites. This leaves us with 120 potential locations.  

In the Future

Next, I’d be interested in bringing in the data of all the locations of houses of worship (and also, potentially, private schools) and doing yet another merge/dissolve to weed out more retail site prospects.