%0 Journal Article
%T 基于机器学习对在线教育用户行为的预测<br>Prediction of Online Education User Behavior Based on Machine Learning
%A 张帅帅
%J Statistics and Applications
%P 225-233
%@ 2325-226X
%D 2022
%I Hans Publishing
%R 10.12677/SA.2022.112024
%X 早在上世纪在线教育就开始在我国崭露头角，发展初期在我国受到各种制约，认可度并不高。然而随着网络不断发展技术不断完善，在线教育发展迅速，目前越来越多人们开始接受在线教育。不只是学生，大学生和工作人群更是在线教育的主要人群。因此网络资源不断增加，各种免费和付费资源层出不穷，很多付费app发现了生财之道，收集有效信息，提高用户对有效知识的接受度。然而，如何找出购买欲望强烈、更有价值的用户，针对性营销，以实现小成本下提升用户转化率是目前互联网普遍面临的问题。本文通过对用户的行为数据进行分析，来挖掘高质量用户所具有的特征，从而帮助企业节省成本，提升利润。针对预处理后的数据集，本文进行了逻辑回归，随机森林预测，XGBoost预测以及LightGBM预测对用户购买行为进行预测，XGBoost以及LightGBM的预测结果相对较好，因此本文是基于XGBoost的预测结果训练和预测的结果对企业提出建议，以提升用户的转化率，增加企业的收入。<br />
As early as the last century, online education began to emerge in my country. In the early stage of development, it was subject to various constraints and the recognition was not high. However, with the continuous development of the Internet, the continuous improvement of technology and the rapid development of online education, more and more people are beginning to accept online education. Not just students, college students and working people are the main population of online education. Therefore, the network resources continue to increase, and various free and paid resources emerge in an endless stream. Many paid apps have discovered the way to make money, collect effective information, and improve users’ acceptance of effective knowledge. However, how to find out more valuable users with strong purchasing desires and target marketing to improve user conversion rate at a low cost is a common problem faced by the Internet at present. This paper analyzes the behavior data of users to mine the characteristics of high-quality users, so as to help enterprises save costs and increase profits. For the preprocessed data set, this paper uses logistic regression, random forest prediction, XGBoost prediction and LightGBM prediction to predict user purchase behavior. The prediction results of XGBoost and LightGBM are relatively good, so this paper is based on the prediction results of XGBoost. The predicted results make recommendations to the enterprise to improve the conversion rate of users and increase the revenue of the enterprise.
%K 用户价值分析，用户转化率，XGBoost模型，特征重要性，机器学习<br>User Value Analysis
%K User Conversion Rate
%K XGBoost Model
%K Feature Importance
%K Machine Learning
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=50023