Прогнозирование повторных продаж с помощью диффузных моделей
Авторы Ду Шаохуэй
Название магистерской диссертации Прогнозирование повторных продаж с помощью диффузных моделей
Факультет Высшая школа менеджмента
Направление подготовки Бизнес-аналитика и большие данные
Год 2021
Научный руководитель Кудрявцев Дмитрий Вячеславович
Описание цели, задач и основных результатов Цель этого исследования – улучшить способность моделей машинного обучения предсказывать поведение потребителей при повторной покупке за счет слияния моделей.
Задачи исследования:
1. Обработка данных и разработка функций.
2. Моделирование и сравнительная оценка моделей.
3. Применение и улучшение модели слияния.
Основные результаты включают систематическое сравнение широко используемых моделей машинного обучения и методов объединения моделей в отрасли, а также усовершенствование метода объединения моделей с накоплением.
Ключевые слова Прогнозирование выкупа, слияние моделей, классификация, классификационная модель машинного обучения.
1.1. Background
With the development of e-commerce in China, major local e-commerce platforms such as Tmall, JD.com, and Vipshop have emerged one after another, and more and more people are participating in online shopping. Compared with traditional shopping models, online shopping has great advantages such as convenience, real-time, and no space restrictions. However, while online shopping brings convenience to people, it also causes problems such as information overload. On the one hand, users are struggling to find their favorite products in the face of a large amount of product information. On the other hand, businesses often get lost in user groups and cannot achieve accurate marketing information delivery.
In order to help users and merchants establish a wide range of contacts, and promote e-commerce platforms to reach more transactions, each e-commerce platform has adopted intelligent systems to analyze users’ needs and preferences and recommend products for them. Consumers’ repurchase behavior prediction is to analyze the their past historical behavior data to analyze the rules of their purchase behavior, and then predict the their future repeat purchase behavior. This technology can be applied to the recommendation system of e-commerce platforms to help merchants identify users with repeated purchase intentions, so that merchants can take more targeted marketing measures, improve user experience, and enable users and merchants to establish a lasting and reliable Relationship, prompting users to buy more products.
1.2. Research motivation
Consumer repurchase behavior is an important measure of their loyalty, and it has always been a research hotspot in the field of marketing. Before the popularization of online shopping, limited by data, the research hotspots of repeat purchase behavior generally focused on its influencing factors, trying to estimate which type of user group would be most likely to produce repeat purchase behavior through the research of influencing factors.
With the development of big data technology and the continuous growth of e-commerce platforms, personal information such as consumer interests and hobbies, and daily shopping behavior information are accumulated in the databases of major e-commerce platforms, gradually forming massive amounts of data. It has been discovered that through the mining of online shopping behavior big data, users’ repeated purchase behavior can be predicted in advance, and it can even be specifically predicted to which merchants each user has repeated purchase intentions. Therefore, the research on repurchase behavior has a new direction. Some studies have begun to pay attention to the technology related to the prediction of user repeat purchase behavior, and gradually apply relevant algorithms in the field of machine learning to the prediction problem of repurchase behavior.
The research of consumer repurchase prediction problem can be applied to the intelligent recommendation system of e-commerce platform to help merchants identify users with repeated purchase intentions. Obviously, a high-precision forecasting system can help more merchants and users establish consumer relationships, thereby bringing huge profits to the e-commerce platform. Therefore, the prediction accuracy and generalization performance of related models used to predict repeat purchase behavior are particularly important.
However, when the existing machine learning model is applied to repeat purchase behavior prediction, its prediction accuracy is low, and the prediction effect is only slightly better than random guessing. Such an effect is difficult to produce valuable effects when applied to a recommendation system. Through combing the literature on the influencing factors of repeat purchase behavior, it is found that consumers’ purchase behavior is controlled by a series of complicated subjective factors, including perceived value, satisfaction, trust, etc. . These influencing factors make the repeat purchase behavior of users diversified and different. When these complex and diverse behavior patterns are hidden in the data, it will bring great difficulties to the fitting of the machine learning model. The difference in user behavior rules is an important reason that restricts the accuracy of prediction, and none of the existing machine learning models can solve this problem well.
1.3. Research goal
Improve the predictive ability of machine learning models on consumer repurchase behavior through model fusion.
1.4. Research question
1. How to mine user data characteristics in multiple dimensions.
2. How to evaluate the prefetching ability of the model fusion method.
3. How to improve the model fusion method.
Academic Literature in English
1.Bauer R A.(1960).Consumer behavior as risk taking.In R.S.Hancock(Ed.), Dynamic marketing for a changing world(389- 398).Chicago: American Marketing Association.
2.Overby J.W., Lee E.J.The effects of utilitarian and hedonic online shopping value onconsumer preference and intentions[J].Journal of Business Research,2006,59(10-11):1160-1166.
3.Chang E.C.,Tseng Y.F.Research note: E-store image, perceived value and perceived risk[J].Journal of Business Research,2013,66(7):864-870.
4.Wang T.G.,Fang Y.H.,Huang H.Y.Understanding customers’repeat purchase intentions in B2C e-commerce:the roles of utilitarian value,hedonic value and perceived risk[J].Information System Journal,2014,24(1):85-114.
5.Teo T.S.H., Wang P, Leong C.H.Understanding Online Shopping Behaviour Using a Transaction Cost Economics Approach[J].International Journal of Internet Marketing &Advertising,2004,1(1):62-84.
6.Mukherjee A,Banerjee S,Bandyopadhyay S.A.Simulation Model Using Transaction Cost Economics to Analyze the Impact of Social Media on Online Shopping[C].ICSI.Springer-Verlag,2012:43-53.
7.Benazić D，Tanković A.Č .Impact of Perceived Risk and Perceived Cost on Trust in the Online Shopping Websites and Customer Repurchase Intention[C].Cromar Congress:Marketing Theory and Practice-Building Bridges and Fostering Collaboration.2015.
8.Myung J.K., Choong K.L., Namho C,et al.Factors Affecting Online Tourism Group Buying and the Moderating Role of Loyalty.[J].Journal of Travel Research,2014,53(3):380-394.
9.Wu L.Y.,Chen K.Y.,Chen P.Y.,et al.Perceived value, transaction cost, and repurchase-intention in online shopping: A relational exchange perspective[J]. Journal of Business Research, 2014,67(1):2768-2776.
10.Yoon C,Gonzalez R,Bechara A,et al.Decision neuroscience and consumer decision making[J].Marketing Letters, 2012, 23(2):473-485.
11.Zhang Y,Feng Y.Q.Factors that Influence a Buyer’s Decision Process of Shopping Online: The Effects of Tradition and Virtual Community[M]. 2011.
12.Tian Y,Ye Z,Yan Y,et al.A practical model to predict the repeat purchasing pattern of consumers in the C2C e-commerce[J]. Electronic Commerce Research, 2015, 15(4):571-583.
13.Chang H.J.,Hung L.P., Ho C.L.An anticipation model of potential customers’purchasing behavior based on clustering analysis and association rules analysis.[J]. Expert Systems with Applications, 2007, 32(3):753-764.
14.Cho Y.S.,Moon S.C., Oh I.B.,et al.Incremental Weighted Mining based on RFM Analysis for Recommending Prediction in u-Commerce[J]. International Journal of Smart Home, 2013,7(6):133-144.
15.Cui D,Curry D.Prediction in Marketing Using the Support Vector Machine[M]. INFORMS,2005.
16.Gupta R,Pathak C.A Machine Learning Framework for Predicting Purchase by Online Customers based on Dynamic Pricing[J]. Procedia Computer Science, 2014, 36:599-605.
17.Zuo Y,Ali A.B.M.S.,Yada K.Consumeer Purchasing Behavior Extraction Using Statistical Learning Theory[J].Procedia Computer Science,2014,35:1464-1473.
18.Hart, P.E. (1968) The condensed nearest neighbor rule. IEEE Transactions on Information Theory, 14, 515-516.
19.D. Shaohui, G. Qiu, H. Mai and H. Yu, “Customer Transaction Fraud Detection Using Random Forest,” 2021 IEEE International Conference on Consumer Electronics and Computer Engineering (ICCECE), 2021, pp. 144-147, doi: 10.1109/ICCECE51280.2021.9342259.
20.Geurts P，Ernst D， Wehenkel L.2006.Extremely randomized trees.Mach Learn，63（1）：3-42.
21.Freund, Yoav; Schapire, Robert E. A Decision-Theoretic Generalization of on-Line Learning and an Application to Boosting. 1995.
22.Chen T，He T，Benesty M，et al．xgboost：Extreme Gradient Boosting [J].2017.
23.Brownlee, Jason (March 31, 2020). “Gradient Boosting with Scikit-Learn, XGBoost, LightGBM, and CatBoost”.
Academic Literature in Other Languages
1.李日旭.网购生鲜农产品中感知风险对购买意愿的影响研究[D].哈尔滨工业大学,2017.
2.黄茜.经济型酒店顾客重复购买行为影响因素实证研究[D].湖北大学,2011.
3.熊晓元.基于互动和感知理论的网络重购行为研究[D].西南交通大学,2014.
4.张延静.网上节日促销中顾客感知价值、顾客满意与顾客网购行为关系研究[D].青岛理工大学,2014.
5.张 春 桃 . 顾 客 感 知 价 值 对 消 费 者 网 购 行 为 影 响 研 究 —— 以 大 学 生 群 体 为 例 [J].商,2016(01):281.
6.肇丹丹.基于 SEM 的消费者网购再惠顾意愿度量分析[J].统计与决策,2014(01):89-92.
7.洪硕.旅游移动电商环境下的游客购买决策影响因素研究[D].广西师范大学,2018.
8.罗胜.电子商务环境下化妆品消费者冲动性购买行为研究[D].华南理工大学,2018.
9.徐雪枫.网络消费者购物决策行为影响因素的实证研究[D].东北财经大学,2012.
10.张春莲.客户购买行为的 BG/NBD 预测模型及其应用研究[D].哈尔滨工业大学,2006.
11.陈洁,谢文昕,杨升荣.在线渠道消费者动态品牌选择购买率预测[J].工业工程与管理,2011,16(03):52-56.
12.舒方,马少辉.客户重复购买的组合预测方法[J].计算机与现代化,2015(05):67-70.
13.李美其,齐佳音.基于购买行为及评论行为的用户购买预测研究[J].北京邮电大学学报(社会科学版),2016,18(04):18-25.
14.杜刚,黄震宇.大数据环境下客户购买行为预测[J].管理现代化,2015,35(01):40-42.
15.刘柏鳞.基于客户网络购物行为分析及预测系统研究[D].北京工业大学,2016.
16.祝歆,刘潇蔓,陈树广,李静,张天宇.基于机器学习融合算法的网络购买行为预测研究[J].统计与信息论坛,2017,32(12):94-100.
17.周成骥.基于机器学习的商品购买行为预测模型设计[D].广州大学,2018.
Последние выполненные заказы
Хочешь уникальную работу?
Больше 3 000 экспертов уже готовы начать работу над твоим проектом!
5 (21 отзыв)
4.2 (25 отзывов)
5 (8 отзывов)
5 (14 отзывов)
4.3 (248 отзывов)
4.1 (20 отзывов)
4.9 (37 отзывов)
5 (18 отзывов)
4.6 (30 отзывов)
