目录
摘要II
关键词II
AbstractIII
引言
1 绪论1
前言 1
1.2 研究内容 1
1.3 研究框架 2
2 隐马尔可夫模型4
2.1 马尔可夫模型 4
2.2 隐马尔可夫模型基本介绍5
2.3 隐马尔可夫模型的三个基本问题7
3 基于隐马尔可夫模型的GDP预测11
3.1 数据预处理11
3.2 建立模型13
3.3 预测算法14
4 总结与展望20
4.1 工作总结 20
4.2 研究展望 20
致谢21
参考文献22
附录1 部分代码（R）23
基于隐马尔可夫模型的GDP预测
摘 要
数据挖掘在各行各业的应用和发展越来越受到人们的关注。随着数据存储的爆炸式增长,机器学习算法将为大数据挖掘带来不可估量的价值。利用机器学习算法分析金融数据可以帮助决策者观察数据特征做出更好的判断。
本文利用隐马尔可夫模型对中国国内生产总值（GDP）进行预测。首先介绍隐马尔可夫模型的形式定义和参数,对于隐马尔可夫模型涉及的三个基本问题进行详细分析了相应的算法。为保证数据的客观有效性，采用官方统计的中国1990至2018年的GDP数据，将前24年数据划分为训练数据，后5年数据作为测试数据，并利用数据求出GDP增长率进行离散化处理。利用离散化后的训练数据进行建模，利用该模型得到最优状态序列，并得到状态序列的转移概率矩阵、观测概率矩阵，由此得到的完整隐马尔可夫模型对后5年GDP增长率进行预测，并将GDP增长率转化为GDP数据。最后，用绝对平均误差（MAD）来评价隐马尔可夫模型下的预测精度。
GDP forecast based on Hidden Markov Model
Student majoring in Statistics Du Shaokang
Tutor Zhang Yibin
ABSTRA *51今日免费论文网|www.51jrft.com +Q: ￥351916072$ 
CT
The application and development of data mining in all walks of life are more and more concerned by people. With the explosive growth of data storage, machine learning algorithm will bring immeasurable value to big data mining. Using machine learning algorithm to analyze financial data can help decision makers to make better judgments in observing the characteristics of data.
In this paper, hidden Markov model is used to predict Chinas GDP. Firstly, the formal definition and parameters of HMM are introduced.Three basic problems related to HMM, and the corresponding algorithm is analyzed in detail. In order to ensure the objective validity of the data, the first 24 years of Chinas GDP data from 1990 to 2018 are divided into training data and the last 5 years are used as test data, and the GDP growth rate is calculated by using the data for discretization. The training data after discretization is used for modeling, and the optimal state sequence is obtained by using the model, and the transition probability matrix and observation probability matrix of state sequence are obtained. The complete hidden Markov model can predict the GDP growth rate in the next five years, and convert the GDP growth rate into GDP data. Finally, the absolute mean error (MAD) is used to evaluate the prediction accuracy under the hidden Markov model.
KEY WORDS: Hidden Markov Model; Forward algorithm; Viterbi algorithm; BaumWelch algorithm; GDP forecast

原文链接：http://www.jxszl.com/jsj/sxtj/606745.html