SVMDecision Tree.docVIP

基于支持向量机和决策树的财经数据分析 摘要 金融市场是国家经济运行的核心，金融时间序列是经济与金融领域中最重要的数据类型，对这类数据进行分析、预测和控制是整个经济和金融活动的重要工作。本文通过支持向量机和决策树两种思想方法对上证指数数据进行分析和预测。 决策树由于算法简单和分类精度高，成为一种广泛应用的归纳推理方法。它无需假设先验概率分布，这种非参数技术具有很好的灵活性和鲁棒性；不仅可以利用离散和连续的数值样本，还可以利用“语义定义”；产生的规则集具有结构简单、容易理解以及计算效率高的特点；可以有效抑制样本噪音和属性缺失问题。 SVM通过核函数对特征空间的映射函数实现非线性到线性的推广。它的优点在于解决小样本，捕捉非线性；克服过度拟合，解决高维问题；解是全局最优，只需要解决二次凸优化；需要设定的参数较少（一般为2~3个），具有很强的灵活性和可拓展性。 关键词：金融时间序列数据 决策树 支持向量机 核函数 Abstract Financial market is the essential economic system of a country.And financial time series is a primary data type in the application of financial area.Analyzing,predicting and controlling of such kind of data is the basic work of the economic and financial activity.SVM ,decision tree are used to analyze and predict the data of SSE index in this paper. Decision tree is one of the most widely used and practical methods for inductive inference because of easily understandable and high classification accuracy.The method has good flexibility and stability,for it could work without prior probability distribution.It can perform not only discrete and continuous data sample but also semantic definition.The rules it generates are simple in structures,easily understood and highly efficiency while calculating.It can effectively suppress sample noise and missing attributes. SVM uses kernel function to extend to nonlinear problems by using its special nonlinear mapping for feature space.It has 4 major advantages:slove small sample problems,capture nonlinear feature;overcome over fitting,solve high dimensions problem;it has best results,what we need to focus on is secondary convex optimization;it need only 2 to 3 parameters,making it high flexibility and expansion. Key words: Financial Time Series Data Decision Tree Support Vector Machine Kernel Function 支持向量机 支持向量机方法是在统计学习理论的基础上发展起来的一种新的机器学习方法。统计学习理论是着重研究有效样本情况下的统计规律以及学习方法的性质，其研究表明经验风险最小化原则下实际风险由经验风险和置信范围两部分组成。经验风险最小化原则在样本有限时是不合理的，统计学习理论在此基础上给出了结构风险最小化归纳原则，旨在针对经验风险和置信范围最小化风险泛函。SVM实现了结构风险最小化原则，保证在有限样