Research Paper: A framework for delineating the scale, extent and characteristics of American retail centre agglomerations

Hey all!

Patrick here from the Geographic Data Science Lab (Liverpool, UK). I am thrilled to present the first paper of my PhD: ‘A framework for delineating the scale, extent and characteristics of American retail centre agglomerations’, available (open-access) at SAGE Journals: Your gateway to world-class journal research. In the paper, using the SafeGraph core places dataset, we extract the spatial extent of retail centres in the Chicago Metro area, and construct a non-hierarchical typology that summarises their key characteristics. We then utilise the centres (and typology) together with weekly patterns data to investigate the response of centres during the early weeks of the COVID-19 pandemic! Check out the figure below :chart_with_upwards_trend:

We are currently working on a follow-up project ‘The Who, What and Where of American Retail Centre Geographies’, which you can check out at: GitHub - patrickballantyne/USRetailCentres: Centres, Characteristics and Catchments of American Retail Centre Agglomerations. :earth_americas:

HUGE thanks to SafeGraph for providing the data, which has been invaluable for my PhD research- you guys are awesome! :safegraph: :100:

Please follow me on twitter, I’d love to connect with you - @pj_ballantyne :bird:


This topic was automatically generated from Slack. You can find the original thread here.

@Patrick_Ballantyne_University_of_Liverpool_UK This is absolutely incredible! I have a few questions.

First- I noticed you used Core and Patterns for this research- did you consider, or make any use of SafeGraph’s Geometry product? Core Places | SafeGraph Docs

Would love your feedback, whether you used it (on its usefulness) or if you took a look and decided not to use it (how might it have been more helpful?)

Second, I wanted to introduce you to @Karen_Ton , whose recent paper, "Customer Loyalty and the Persistence of Revenues and Earnings", was also examining retail centers using SafeGraph data.

@Karen_Ton - Patrick’s project, “The Who, What and Where of American Retail Centre Geographies”, is developing a comprehensive understanding of the geographies of American retail centers. Do you know of any resources he might want to add to this page? GitHub - patrickballantyne/USRetailCentres: Centres, Characteristics and Catchments of American Retail Centre Agglomerations

Last, @Patrick_Ballantyne_University_of_Liverpool_UK - I noticed you used H3 for classification. Do you know much about Placekey? If not, I think you’d be really interested: How it Works

Basically, it’s a free, universal location identifier generated by a group of partner organizations who wanted to make it easier to join and share data across datasets and organizations. The Placekey’s characters are actually based on H3! All of SafeGraph’s data is Placekeyed, which makes it easier to join to other datasets. There’s an interesting webinar here that talks about how you can use Placekey’d datasets like SafeGraph’s, and zoom in and out on the different levels of resolution with H3 in order to conduct interesting research, like correlating neighborhood demographics with a retail store’s purchasing patterns: Correlating Neighborhood Demographics and Retail Purchasing Patterns Using Free Tools. - YouTube

@Patrick_Ballantyne_University_of_Liverpool_UK Really enjoyed reading this paper and glad SafeGraph Data was helpful in your research.

I think you may also like this paper by other members of the community. One of the findings was, “comparison-good stores relates to higher trip likelihoods and more customer visits per cluster” discussed in section 3.3. It may provide more insight into another region besides Chicago and introduce you to some different methodology.

At the end of the paper you discuss a few different applications for the framework you and your team discussed. Is there one that you are most excited about?

Congratulations @Patrick_Ballantyne_University_of_Liverpool_UK ! What a great way to kick off your PhD journey! I’m sure this will be one of many publications to come. I’m especially excited to learn more about your follow-up project. Be sure to keep us in the loop!

Loved the citation to the work of some of our other Community members. One of the cool perks of being in the Community is that you can connect with the authors of papers that you cited. Going to loop them into our conversation.

Hey @Song_Gao_UW-Madison and @Yuhao_Kang_University_of_Wisconsin! Did you see this publication by Patrick?

It cited your team’s paper, Association of Mobile Phone Location Data Indications of Travel and Stay-at-Home Mandates With COVID-19 Infection Rates in the US. Pretty neat! I’ll link to their paper here if other members are wanting to check it out too. Very cool to see the same underlying data used in a wide variety of applications, from COVID-19 to retail centers!

This work also reminds me of another paper by a few Community members, @Avi_Goldfarb_University_of_Toronto and @Catherine_Tucker looking at which retail outlets generate the most physical interactions. Have you already checked this publication out? Here’s a link to the working paper.

Hey @Avi_Goldfarb_University_of_Toronto and @Catherine_Tucker ! Have you checked out this publication by Patrick? I know your work focused more broadly in the US, not necessarily a specific city. How do you think these findings might change depending on the type of city (i.e., rural vs urban, cities experiencing different weather, etc.)?

I also saw that you used employment data from the US Census. Thought I might share that we’ll have a third party dataset available in Shop in mid-October from Netwise. This will be another free dataset available in Shop that is exclusively employee data. Might be helpful to keep in mind as you pursue further research that might utilize employee data.

As you continue your work on creating definitions for Global Retail Centers, you might also keep an eye on an upcoming event we have. We’ll be having an event announcing our Global Brands product! Think Core Places but for the entire world! Might be helpful in your future research. Here’s more info if you’re interested: Sign up for Going Global: A Webinar Announcing SafeGraph Global Brands

Again, congrats and welcome to the Community! :partying_face:

Thanks for sharing this great research!

Congratulations to @Patrick_Ballantyne_University_of_Liverpool_UK ! And please say hi to Alex!

Hi @Karissa_from_SafeGraph ! It’s lovely to meet you… Very excited to receive some community feedback, let me try and give you my thoughts on some of your questions…

  1. Yes, I have in fact been using the building geometries comprehensively in my current project - as you can see in the paper the centre boundaries are slightly inaccurate due to use of points over polygons in boundary delineation - the polygons (supplemented with land-use polygons from OpenStreetMap) have proved invaluable in identifying agglomerations of retail places. My feedback on the geometry product is only positive - very easy process of access, very easy to use and highly accurate! Thankyou SafeGraph!!
  2. Hey @Karen_Ton ! Nice to meet you, we should connect and consolidate our research at some stage - I am on twitter (@pj_ballantyne)
  3. Yes, I am familiar with placekey, and the idea to use these aggregate geometries did cross my mind. However, as a lab we have been developing a methodology using graph structures and H3 geometries to build UK retail centre definitions Retail Centre Boundaries | CDRC Data, so we wanted to provide some consistency between definitions/methodologies, for a cross-regional comparison as part of my third paper!

Hey @Hayden_SafeGraph , it’s great to meet you! Really pleased to hear you enjoyed reading the paper - hugely appreciated.

Thanks for suggesting the community paper, i’ll add it to my list :slightly_smiling_face:

Regarding the future directions of my research, we have actually explored most of the suggestions we made in my (upcoming) second paper, which is super exciting! We developed a computationally stable methodology for retail centre delineation using H3 (USRetailCentres/Helper Functions - Delineation (Parallel).R at main · patrickballantyne/USRetailCentres · GitHub), we have expanded the typology to the c.11,000 centres delineated across the US, and incorporated ‘better’ variables, such as premium brands, anchors, measures of urban morphology and local/national diversity indices (USRetailCentres/Helper Functions - Typology (Parallel).R at main · patrickballantyne/USRetailCentres · GitHub) - the typology looks really great, more on this soon!!

Thanks for the feedback, happy to chat more if interested! :smile:

@Niki_from_SafeGraph Thanks a lot for linking these useful papers/people from the community, it’s great to see the network of people involved in similar stuff! :slightly_smiling_face:

I will be sure to check out the SafeGraph global brands event! My thesis has predominantly focused on US retail centres (& their geographies), but I know a lot of people in our research cluster that will be very interested in this!

Hey @Song_Gao_UW-Madison , it’s great to meet you! It was actually at your presentation of your paper on calibrating the dynamic huff model (https://arxiv.org/pdf/2003.10857.pdf) that I was first introduced to SafeGraph - so thanks a lot!

Recently we have been working on retail centre catchments using the approach in that paper, using the SafeGraph weekly patterns data to calibrate Huff model parameters and estimate catchments - the paper was awesome and provided a lot of inspiration, so thanks again!

Hope to meet you in person in the near future - I will send regards to Alex too :smile: