Abstract:

Aiming at the problem of irregular suspension and long suspension of stock, a combination model of stock suspension prediction is proposed by using machine learning technology. The financial and stock indicators are selected as data, and divided into multiple feature data sets by calculating the importance of indicators. Multiple classification models are built to form a model pool, from which system randomly extracts some of models for classification, and obtain the final prediction result by voting method. Empirical analysis takes listed companies in China as the research object. The results of the experiments show that the proposed system has relatively high accuracy and can reduce misdiagnosis rate and omissive judgement rate compared single model.

Key words: machine learning, stock suspension, combination model, random forest

摘要：

针对股票的无规律性停牌和长时间停牌的问题，采用机器学习相关技术，提出了股票停牌预测的组合模型，选取财务和股票两方面的指标作为数据，通过计算各个指标的重要性进行筛选，并划分出多个特征数据集，进而完成多个分类子模型的学习，形成子模型池，系统随机抽取多个子模型进行分类，并通过投票法得到最终的预测结果。实证分析以中国上市公司为研究对象，结果表明组合模型预测取得了较高的准确率，与单一模型相比在误报率和漏报率上有较大的改进。

关键词: 机器学习, 股票停牌, 组合模型, 随机森林