English |
讲座题目：Unveiling Consumer Preference from Filtering Choices Using a Bayesian Dynamic Approach
讲座时间：2023年10月11日10:30-12:00
讲座形式：腾讯会议（会议号：396-894-100）
主持人：李凯教授
特邀主讲嘉宾介绍：
董晓静老师现任圣克拉拉大学利维色情导航 营销系教授。在此之前，她在清华大学获取学士学位，在麻省理工学院（MIT）获取硕士学位，在美国西北大学（Northwestern University）获得博士学位。董教授的研究方向为人工智能和大数据的方法及其在市场营销和商业决策中的应用，侧重于对消费者决策过程和变化的分析和研究。她的主要研究课题是商业行为如何影响客户的行为和决策以及客户之间的互相影响，包括产品销量预测、用户决策动态分析、计量经济学、机器学习和贝叶斯统计学。董教授的多篇文章发表在UTD期刊，如Marketing Science，Journal of Marketing Research等，并多次得到媒体报道和引用。
讲座内容介绍：
Making inferences about consumer preferences has been instrumental in targeting and personalized recommendations. Traditional methods have relied on access to historical consumer-level data, a resource becoming scarce due to burgeoning privacy regulations. In this research, we introduce a Bayesian dynamic approach to decode consumer preferences by harnessing their search filtering choices such as adjustments on price ranges to refine search results. This innovative method allows us to understand consumer preferences even with limited data. This understanding holds the potential to bolster a company’s capability to provide timely recommendations while adhering to the constraints of privacy regulations. We develop a Bayesian model to capture live-streamed information to assess individual-level price-quality tradeoffs in a fluid setting. Through a utility model tailored for price-quality tradeoffs, our model provides a closed-form solution, which brings a transparent insight into how various factors converge to influence consumer choices. We validate the model’s efficacy through simulations and actual data from a prominent travel agency, deploying the Markov chain Monte Carlo technique. Both simulated and real-world applications of this novel method underscore its superior predictive prowess, suggesting that businesses can adeptly discern consumer inclinations by analyzing their real-time online actions.
