随着大数据产业的高速发展，用户维度的大数据在金融机构的经营决策中发挥着越来越重要的作用。互联网金融借助其强大的数据获取和处理能力，使得用户大数据作为决策参考的支持，在包括信用评估、风险管理、精准营销等多个环节内逐步成为主流。然而，在学术界研究中，对用户维度金融大数据应用的定量研究在互联网金融领域还留有一定行业空白，尤其是互联网金融领域的精准营销方向尚缺少从数据出发的量化分析。因此本文在分析了互联网金融模式的基础上，基于互联网金融平台X的用户数据，利用机器学习算法，在用户分层的基础上对用户进行聚类分析，根据不同聚类用户的特点和互联网金融零售的业务属性，充分挖掘用户价值和投资需求，并基于此制定了个性化的用户精准营销策略，以期达到投资用户转化、客户留存提升、流失风险控制的业务目标。具体基于用户画像的精准营销研究内容如下：

比较研究了机器学习中的聚类算法，根据有无投资行为将用户分为两类，选取用户基础属性、社交需求和金融偏好数据作为变量，利用基于划分的K-means算法分别对两类用户进行聚类分析，根据聚类结果为不同类型的用户进行特征倾向画像，并基于画像结果制定个性化的营销和维护策略。

With the rapid development of big data industry, user-dimensional big data is playing an increasingly important role in financial institutions' business decision making. With the powerful data-acquisition and processing capabilities, the industry of Internet Finance makes it a mainstream that big data works as a support for decision-making in product
pricing, credit evaluation, risk management and precise marketing. However, in the academic world, quantitative research on the application of user-dimensional financial big data remains quite blank in the field of Internet Finance. In particular, quantitative analysis based on big data is largely needed in the area of precise marketing and user portraits. Based on the analysis of the Internet Financial model, this paper includes user data of Internet Finance platform X, applying machine learning algorithms to conduct the hierarchical clustering analysis. According to the characteristics of different clusters and the attribute of Internet Financial Retail business, the paper fully mines business value and investment demand of users, developing personalized and precise marketing strategies to achieve the transformation of investment user, customer retention improvement and loss-risk control. The specific research content of precise marketing based on user portraits is as follows:

Based on the investment behaviors, users can be divided into two categories. The paper conducts the comparative study on clustering algorithms of machine learning, selects users’ fundamental character,
social demand and financial preference data as variables, applying the K-means algorithm for the cluster analysis of two types of users. According to the clustering results, the paper conducts user portrait, developing personalized marketing and maintenance strategies based on the user portraits.