Research on predicting the impact of promotional activities on consumer behavior in omnichannel retailing

Abstract

In recent years, the omnichannel sales model has rapidly emerged and is widely used in a context in which consumer buying behavior is influenced by a variety of factors that produce complex changes. Among these factors, the impact of omnichannel promotions on consumer behavior is particularly critical, and thus this research area has received much attention. Predicting the shopping feedback of different types of consumers with different numbers of promotions can help companies develop targeted promotional strategies and improve the consumer shopping experience. In this paper, we classify consumers into different types by clustering them in three dimensions: basic personal information, shopping preference, and shopping channel preference. Based on this, a machine learning prediction model based on the random forest algorithm is built in each of the three dimensions, and the model's performance is evaluated by plotting ROC curves. In order to improve the model's performance, this paper uses the up-and-down sampling method to balance the data. The prediction model based on consumers' "basic personal information", "shopping preference," and "shopping channel preference" has been successfully developed, and the prediction results are excellent. This study provides guidance for companies to develop targeted promotion strategies and offers new ideas and methods for marketing and data analysis. In the future, we can continue to dig deeper into consumer behavior data, classify consumers in more dimensions, build more accurate prediction models, and continue to improve the scientific and accurate promotion decisions of enterprises.