Precise Issuance of Meituan Merchants’ Coupons with Machine Learning

Abstract

With the popularity of mobile Internet, the “Online-to-Offline” (O2O) business model has become popular. Issuing coupons to attract new customer registrations and keep old customers active is an important marketing tool for O2O companies. But the random distribution of coupons can be annoying to those non-target customers. For merchants, the transition of issuing coupons to merchants will not only increase the promotion cost but also have a negative effect on their brand reputation. The purpose of this study is to analyze transaction data and build a model to predict the redemption of coupons, so as to achieve the precise issue of coupons by merchants. We use machine learning to analyze the consumption data and extract features from five categories: coupons, merchants, consumers, consumers-merchants, and other categories. A total of 44 features are extracted and the XGBoost (eXtreme Gradient Boosting) model is adopted. It has been verified that the prediction results of the application of the XGBoost model can nearly increase 50% net profits of the merchants.