Predicting Global Restaurant Facility Closures

57 Pages Posted: 18 Jul 2019 Last revised: 26 Mar 2020

See all articles by Derek Snow

Derek Snow

The Alan Turing Institute

Date Written: January 23, 2018


This paper predicts the likelihood that a restaurant will close within the next one to two years using a Yelp restaurant dataset and a high dimensional gradient boosting machine called LightGBM (hereafter GBM). This model, trained on more than 20,000 individual restaurants, has an accuracy just above 96% and an ROC (AUC) score of 75%. An ROC (AUC) score above 70% is ordinarily classified as a “fair model” in terms of performance. Using the prediction model, I also quantify the most predictive variables and higher-order variable interactions, both of which produce compelling insights into several non-linear relationships. A model that predicts facility closures has implications for both equity and debt providers. In this chapter, I argue that capital providers should make use of publicly available datasets to aid their capital allocation decision-making process.

Keywords: Machine Learning, Applied, FirmAI, Restaurant, Bankruptcy, Failure, Closures

JEL Classification: C38, C45, C52, C53, C54

Suggested Citation

Snow, Derek, Predicting Global Restaurant Facility Closures (January 23, 2018). Available at SSRN: or

Derek Snow (Contact Author)

The Alan Turing Institute ( email )

British Library, 96 Euston Rd
London, NW1 2DB
United Kingdom


Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics