Americans are spending a larger amount of their income on eating at restaurants than ever before. The National Restaurant Association estimates 2019 sales at roughly $863 billion almost double the $590 billion in 2010. As a result, many technological review services have arisen to help these eager consumers find their best possible dining experience; whether that be determined by cost, menu options, or otherwise. As such, our research seeks to answer if restaurant review technology data can be used as a powerful indicator of future restaurant success. Our research discusses a novel solution to predicting restaurant success through the creation of a comprehensive dataset and the use of various machine learning models. Existing literature that discusses measuring restaurant success focuses on restaurant-specific data to form their independent variables. Our research demonstrates that community-specific data and density data expressing the uniqueness of a restaurant contribute important information to predicting restaurant success. It also shows that year 1 and year 2 review counts are strong indicators of year 3 review counts, hinting at the importance of a ramp-up period.
This project uses machine learning and statistical analysis to predict the success of a restuarant.
You can read more on Medium: https://medium.com/analytics-vidhya/predicting-restaurant-success-based-on-review-technology-and-zip-code-level-demographic-data-ba48ca53619f
File Descriptions:
Modeling Notebook: Includes code for both regression and neural network models
Data Combination Notebook: Combines various datasets
Dataset descriptions: