|dc.description.abstracteng||The proliferation of mobile devices especially smart phones brings remarkable opportunities for both industry and academia. In particular, the massive data generated from users' usage logs provide the possibilities for stakeholders to know better about consumer behaviors with the aid of data mining. In addition, with the popularization of the mobile Internet and the prevalence of delivery service, Online Takeout Ordering & Delivery (OTOD) using Apps from smart phones or websites from PC has become an emerging service and prosperous industry(e.g., KFC delivery). Merchants sometimes run big promotions (e.g., discounts or cash coupons) on particular dates (e.g., Boxing-day Sales, "Black Friday" or "Double 11 (Nov 11th)", in order to attract a large number of new buyers. Unfortunately, many of the attracted buyers are one-time deal hunters, and these promotions may have little long lasting impact on sales. To alleviate this problem, it is important for merchants to identify who can be converted into repeated buyers. By targeting on these potential loyal customers, merchants can greatly reduce the promotion cost and enhance the return on investment (ROI).
Firstly, we studied the consumers' short-term and long term behavior across different platforms comprehensively. Then we tried to find a series of features to deal with the problem of repeat buyer prediction in E-commerce.
For the consumer behavior analysis, we examine the consumer behaviors across multiple platforms based on a large-scale mobile Internet dataset from a major telecom operator, which covers 9.8 million users from two regions among which 1.4 million users have visited e-commerce platforms within one week of our study. We make several interesting observations and examine users' cultural differences from different regions. Our analysis shows among the multiple e-commerce platforms available, most mobile users are loyal to their favorable sites and people (60%) tend to make quick decisions to buy something online, which usually takes less than half an hour. Furthermore, we find that people in residential areas are much easier to perform purchases than in business districts and more purchases take place during non-work time. Meanwhile, people with medium socioeconomic status like browsing and purchasing on e-commerce platforms, while people with high and low socioeconomic status are much easier to conduct purchases online directly. We also show the predictability of cross-platform shopping behaviors with extensive experiments on the basis of our observed data.
In order to improve the quality of service and recommendation personalization, we tried to find the key factors leading to a successful purchasing of takeout food in this paper. We collected Internet access records related to OTOD service of 34,845 users with a time duration of nearly four months. At first, we did a preliminary study on users' daily and periodic purchasing behaviors of takeout food. Then we combine the demographic information and location information with the purchasing activities to find the most potential purchasing groups of takeout food. Based on the features extracted from historical purchasing records, demographic information and location information, we use several popular machine learning methods to predict the future purchasing activities within a specific time. The experiments show that our extracted features can be well used for the takeout food purchasing prediction problem.
It is well known that in the field of online advertising, customer targeting is extremely challenging, especially for fresh buyers. With the long-term user behavior log accumulated by Tmall.com, we get a set of merchants and their corresponding new buyers acquired during the promotion on the "Double 11" day. Our goal is to predict which new buyers for given merchants will become loyal customers in the future. To achieve this goal, we did a comprehensive feature engineering to find the key factors influencing consumers' repeat purchasing in the future. Based on the features, we build a merged machine learning model to predict the repeat buyer and achieve a roc-auc score with 0.697.||de