基于K-Prototypes聚类算法的股票分析师行为划分

The Behavior Analysis of Stock Analysts Based on K-Prototypes Clustering Algorithm

摘要: 股票分析师作为信息中介，通过发布研报的形式提供股票内在投资价值的信息，其行为越发受到广大投资者的关注。由于股票分析师数量众多、研报风格迥异、质量良莠不齐，投资者缺乏相关知识经验难以去选择适合自身偏好的分析师研报。本文利用K-prototypes聚类算法分析具有混合属性的股票分析师行为数据，解决了股票分析师群体数据量大且分散的特性。通过刻画不同股票分析师群体的特征，帮助投资者了解分析师群体获取更多有价值的数据信息，进行理性投资降低投资风险，同时其结果为后续的多元分析提供数据基础。

Abstract: As the information intermediary, stock analysts provide information about the inner investment value of the stock by publishing research report and their behavior is more and more concerned by the investors. Because of the large number of stock analysts, the different style and quality of re-search report, the investors lacking the relevant knowledge experience have difficulty in choosing the research reports which are suitable for their own preference. This paper uses the K-prototypes clustering algorithm to analyze the behavior of stock analysts with mixed attributes, which solves the large and the dispersive characteristics of the stock analysts’ group. By depicting the characteristics of different stock analysts’ group, investors can know the analysts’ group better to obtain more valuable data information and make investment rationally to reduce the risk of investment. The results can also provide the data basis for the follow-up multivariate analysis.