User Behavior Analytics and Software Development for Assessing RaceRunner Customers****

留住客户对公司的发展至关重要。如果你不能留住新用户，那么无论你花多少钱去获取新用户都无济于事。在这个项目中，我们将制定策略来模拟复杂的用户行为，从而使用贝叶斯网络、预测模型和机器学习算法来更深入地了解应用程序用户客户。贝叶斯方法允许我们根据数据告诉我们的情况，对这些预测模型应该是什么样子的先验信念进行编码。这些方法将允许根据他/她过去的数据预测离开任何新客户或现有客户的可能性。这对于根据客户的行为识别即将离开服务的客户并采取适当的补救措施非常重要。这些方法还将促进客户的行为聚类和用户流失预测，以便更好地了解业务的健康状况。这些方法将在RaceRunner客户提供的100多万条结构化、非结构化和多维数据记录上进行测试。然后，我们的目标是开发一个软件平台，使它可以为没有任何分析知识的决策者和客户使用。该软件将允许公司分析和预测他们的客户选择行为，并促进更好地理解这种行为，这将允许RaceRunner引入更多成功的功能和服务，并增加他们的市场份额****

Customer retention is most important to the growth of a company. It doesn't matter how much you spend to acquire new users if you can't retain them. In this project, we will develop strategies to model complex user behaviors to gain a deeper understanding of app user customers using Bayesian networks, predictive models and machine learning algorithms. Bayesian methods allow us to encode our prior beliefs about what these predictive models should look like, conditioning on what the data tell us. These methods will allow to predict the probability of leaving any new or existing customer given his/her past data. This is important for identifying the customers who are about to leave the service, based on their actions and take appropriate remedial measures. These methods will also facilitate the behavioral clustering of customers and user churn prediction in order to better understand the health of the business. The methods will be tested on over one million records of structured, unstructured and multidimensional data available from RaceRunner customers. We then aim to develop a software platform to make it usable for decision makers and customers without any analytics knowledge. The software will allow the company to analyze and predict their customer choice behavior and facilitate a better understanding of this behavior which will allow RaceRunner to introduce more successful features and services and increase their market share.****