﻿ 中国房地产价格的研究—基于ε-TSVR模型和VAR模型 Research on the Price of Real Estate in China—Based on ε-TSVR Model and VAR Model

Statistical and Application
Vol.04 No.03(2015), Article ID:16057,12 pages
10.12677/SA.2015.43022

Research on the Price of Real Estate in China

—Based on e-TSVR Model and VAR Model

Lingling Xie, Yu Zhang

School of Statistics and Mathematics, Yunnan University of Finance and Economics, Kunming Yunnan

Email: xielingling1992@163.com

Received: Aug. 29th, 2015; accepted: Sep. 12th, 2015; published: Sep. 18th, 2015

ABSTRACT

China’s price of real estate forecasts has been a hot livelihood issue, and scholars have paid much attention to it. In this paper, the monthly data of the national 2005-2013 years, in the study of China’s real estate prices, are based on the use of vector auto regression VAR model and support vector regression (e-TSVR) model, respectively to predict and compare the Chinese real estate prices. The results show that the average absolute error (MAE), the average absolute percentage error (MPE), the root mean square error (RMSE) value of the of e-TSVR model are less than VAR model, which shows that the e-TSVR model has better forecasting effects on the real estate prices in China.

Keywords:Real Estate Price, VAR Model, e-TSVR Model

—基于e-TSVR模型和VAR模型

Email: xielingling1992@163.com

1. 引言

2. 房地产价格影响因素

3. 模型介绍

3.1. e-双子支持向量回归(e-TSVR)

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

3.2. VAR模型

(29)

(30)

(31)

(32)

VAR模型虽然看着比一般模型较为复杂，但是它的求解却不繁琐。常见的，VAR模型估计求解方法有最大似然估计(MLE)和最小二乘估计(OLS)，最大似然估计和最小二乘估计求解得到的各参数是一致的。

4. 实证分析

4.1. 相关数据说明

4.2. e-TSVR模型拟合

e-TSVR模型的预测本文选取的核函数为高斯核函数：

SSE (误差平方和)：

SST (总平方和)：

SSR (回归平方和)：

NMSE (正则均方误差)：

(决定系数)：

4.3. e-TSVR模型预测及结果分析

4.4. VAR模型拟合

Figure 1. Price data of actual value and the fitting value comparison of e-TSVR model

Figure 2. Price data of actual value and the predictive value comparison of e-TSVR model

Figure 3. Price data of actual value and the fitting value comparison of VAR model

Table 1. Calculating value of the performance formula of fitting accuracy

Table 2. Calculating value of the performance formula of the forecast accuracy

Table 3. Calculating value of the performance formula of fitting accuracy

4.5. VAR模型的检验

4.6. VAR模型预测及结果分析

5. 结论

5.1. e-TSVR模型与VAR模型的比较分析

MAE (平均绝对误差)：

MPE (平均绝对百分误差)：

RMSE (均方根误差)：

Figure 4. AR root graph

Figure 5. Price data of actual value and the predictive value comparison of VAR model

Figure 6. Comparison between the actual value and the predictive value of monthly house price based on VAR model and e-TSVR model in 2013

Table 4. Results of Granger causality test

Table 5. Calculating value of the performance formula of fitting accuracy

Table 6. The results of the indicators of e-TSVR model and VAR model

5.2. 总结

Research on the Price of Real Estate in China—Based on ε-TSVR Model and VAR Model[J]. 统计学与应用, 2015, 04(03): 196-207. http://dx.doi.org/10.12677/SA.2015.43022

1. 1. 王彬 (2007) 房地产价格影响因素分析. 硕士论文, 北京交通大学, 北京.

2. 2. 罗玉波 (2011) 房价影响因素分析:分位数回归方法. 统计与决策, 6, 158-159.

3. 3. 李勇, 王有贵 (2011) 基于状态空间模型的中国房价变动的影响因素研究. 南方经济, 2, 38-45.

4. 4. 李芳, 李秋娟 (2014) 人民币汇率与房地产价格的互动关系. 国际金融研究, 3, 86-96.

5. 5. 李成, 马国校 (2007) VAR模型在我国银行同业拆借市场中的应用研究. 金融研究, 5, 62-77.

6. 6. 张卫平 (2012) 中国通货膨胀预测：基于AR和VAR模型的比较. 统计与决策, 4, 11-15.

7. 7. 刘晓曙, 郑振龙 (2007) 商业银行VAR 模型预测能力的验证. 当代财经, 8, 39-43.

8. 8. 彭显刚, 王洪森 (2014) 基于竞争ISPO双胞支持向量回归短期负荷预测. 电力系统及其自动化学报, 10, 46-68.

9. 9. 曹慧 (2014) 基于支持向量回归机的中国物价波动影响因素探究. 硕士论文, 浙江工业大学, 杭州.

10. 10. 鲍漪澜 (2013) 基于支持向量机的金融时间序列分析预测算法研究. 博士论文, 大连海事大学, 大连.