统计与数据科学学院知识库

随着互联网与大数据技术愈益成熟以及我国对外贸易广度和深度不断加强，再加上电子商务与国际贸易活动深度融合，跨境电子商务应运而生并快速发展。统计资料显示，2009年我国跨境电子商务交易总额仅为0.9万亿元，2014年上升为4.2万亿元，而到2019年其交易规模已突破10万亿元，同时，跨境电子商务的迅速发展使其占我国进出口总额的比重也从2009年的6%提高至2014年的15.9%，到2019年其占比已高达33%。可以看出，在我国传统外贸增速放缓的背景下，跨境电子商务作为电子商务的最新发展领域，不仅保持了快速增长态势，而且已成为我国对外贸易发展的重要增长点。然而，随着逆全球化浪潮不断升级，跨境电子商务行业竞争也愈发激烈，在面临行业标准有待完善、政策法规体系不健全、配套服务不成熟、专业人才匮乏、跨境物流运输效率低等诸多发展困境的同时，贸易成本不断增加以及消费市场细分要求进一步提高也使其持续快速发展受到极大制约。因而，在此背景下，为进一步提高我国跨境电子商务营销水平和市场占有率，本文以JDC集团跨境电子商务企业为例，从用户视角出发，充分利用JDC集团跨境电子商务平台交易数据，按照理论研究——现状分析——目标用户定位——预测模型构建——营销建议提出的研究思路，对JDC集团跨境电子商务平台营销策略进行探究。文章首先对数据库营销和跨境电子商务理论研究成果进行阐述，并基于平台基础指标数据对JDC集团跨境电子商务平台发展现状进行分析总结，接着依据现状分析结果进一步归纳平台在产品、渠道、客诉及商家管理、数据四个层面存在的主要问题，然后从用户基本特征、用户价值、用户偏好三个角度出发，全面探究平台目标用户的精准化定位，在总结平台目标用户特征的基础上，得出目前平台用户忠诚度偏低、消费能力处于中下等水平、低价值用户占比大以及不同性别、年龄、地区的用户偏好差异较大等结论，进而以GBDT机器学习算法概述及模型效果评价指标说明为基础，依照数据预处理——样本选取——特征变量构建与选择——模型训练的流程构建用户购买行为预测模型，并得到F1值为92.1%、AUC的值为0.965，模型预测效果较好的结论。最后根据前文分析结果首先从巩固平台已有用户视角提出品类优化、品牌推广、用户个性化推荐、跨境物流服务等方面的营销建议；接着又从拓展新用户视角以产品、市场、运营三方面为出发点提出平台拉新引流的相应策略。

As the Internet and big data technology become more and more mature, the breadth and depth of China's foreign trade continue to strengthen, coupled with the in-depth integration of e-commerce and international trade activities, cross-border e-commerce arises at the right moment and develops rapidly. Statistics show that the total amount of cross-border e-commerce transactions in China was only 0.9 trillion yuan in 2009, rising to 4.2 trillion yuan in 2014, and by 2019, the transaction size has exceeded 10 trillion yuan. Meanwhile, The rapid development of cross-border e-commerce has increased its share of China's total import and export volume from 6% in 2009 to 15.9% in 2014 and up to 33% in 2019. It can be seen that in the context of China's traditional foreign trade growth slowing down, cross-border e-commerce, as the latest development field of e-commerce, has not only maintained a rapid growth trend, but also become an important growth point of China's foreign trade development. However, as the tide of anti-globalization continues to escalate, the competition in the cross-border e-commerce industry is becoming increasingly fierce. In the face of many development difficulties, such as unperfected industrial standards, imperfect policies and regulations, immature supporting services, lack of professionals, and low efficiency of cross-border logistics and transportation, The continuous increase of trade costs and the further improvement of consumer market segmentation requirements also make its sustained and rapid development is greatly restricted. Therefore, in this context, in order to further improve the level of cross-border e-commerce marketing and market share in China, this paper takes the cross-border e-commerce enterprise of JDC Group as an example, and makes full use of the transaction data of the cross-border e-commerce platform of JDC Group from the perspective of users. According to the research ideas of theoretical research, current situation analysis, target user positioning, prediction model construction and marketing suggestions, the marketing strategy of JDC cross-border e-commerce platform is explored. This paper on database marketing and cross-border e-commerce elaborates the theoretical research results, and data index is based on the platform of JDC group cross-border e-commerce platform development present situation carries on the analysis summary, then according to the status quo analysis results further induces platform in the product, channel, customer complaint and business management, main problems of data of four levels, Basic features from a user, the user value, then the user preference three perspective, comprehensive inquiry platform target users precise positioning, on the basis of summarizing the platform target user characteristics, it is concluded that the current platform user loyalty is low, consumption ability is in the lowest level, low value accounted for big users and user preferences of different gender, age, area differences such as conclusion, Then, based on the GBDT machine learning algorithm outline and the model effect evaluation index description, according to the process of data preprocessing -- sample selection -- characteristic variable construction and selection -- model training, the user purchase behavior prediction model is built, and the F1 value is 92.1%, the AUC value is 0.965, the model prediction effect is good. Finally, according to the above analysis results, the marketing suggestions on category optimization, brand promotion, personalized recommendation of users, cross-border logistics services and other aspects are proposed from the perspective of consolidating the existing users of the platform. Then from the perspective of expanding new users from the product, market, operation as the starting point of three aspects of the platform to draw new drainage of the corresponding strategy.