Machine Learning Techniques for Prediction of Retail Violation of Tobacco Products

• Activity Number: 499 - New Methods for Machine Learning
• Type: Contributed
• Date/Time: Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Adams Kusi Appiah* and Hongying Dai
• Companies: University of Nebraska Medical Center and University of Nebraska Medical Center
• Keywords: Tobacco; machine learning ; prediction ; retail violation

Abstract

Effective on August 6, 2016, the finalized “Deeming Rule” extended the Food and Drug Administration (FDA) authority to regulate e-cigarettes, cigars, and other newly deemed tobacco products. We evaluate how machine learning techniques can be applied to predict retail violations by tobacco products and examine the neighborhood characteristics associated with retail violations of sales to minors (RVSM) using inspection data on tobacco retailers from the FDA compliance check database. We will train different machine learning algorithms to predict retail violations and performed a comprehensive comparison of the different techniques used to find the best performance. Some examples of the techniques are Random Forest, Neural Networks (NN), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), etc. A conclusion of the best technique will also be provided.