﻿ 京沪深房地产上市公司的盈利能力分析——基于随机森林 Analysis of Profitability of Real Estate Listed Companies in Beijing, Shanghai and Shenzhen—Based on Random Forest

Finance
Vol. 09  No. 02 ( 2019 ), Article ID: 29341 , 8 pages
10.12677/FIN.2019.92011

Analysis of Profitability of Real Estate Listed Companies in Beijing, Shanghai and Shenzhen

—Based on Random Forest

Junqi Guan, Lin Li

School of Mathematics and Systems Science, Shandong University of Science and Technology, Qingdao Shandong

Received: Feb. 26th, 2019; accepted: Mar. 14th, 2019; published: Mar. 21st, 2019

ABSTRACT

According to the development of Beijing, Shanghai and Shenzhen real estate companies, choosing appropriate profitability indicators, the 57 listed real estate companies of Beijing, Shanghai and Shenzhen are classified, and the Random Forest model is established. The results show that the number of profit rankings of real estate listed companies in Beijing, Shanghai and Shenzhen is balanced, and the model by random forest has a high precision, which can effectively predict the profitability level of real estate listed companies in Beijing, Shanghai and Shenzhen. Based on the above conclusions, in response to the development of China’s real estate industry, the corresponding policy recommendations for the future growth of the industry and how real estate companies improve their profitability are proposed.

Keywords:Real Estate, Listed Company, Profitability, Random Forest

——基于随机森林

1. 引言

2. 理论阐述

1) 输入参数：原始训练集样本的个数 $N$ ，变量的数目 $M$ ，节点分类变量 $m$

2) 用自助抽样法有放回地随机抽取 $k$ 个自助样本集，根据自助样本集，构建 $k$ 棵决策树，每次未被抽到的样本组成了 $k$ 个袋外数据，即简称OOB；

3) 在训练的过程中，自助样本集逐渐生成一棵棵决策树，在每个节点处，随机选取 $M$ 个特征值中的 $m$ 个( $m$ 小于 $M$ )，随后从 $m$ 个特征中选取一个特征进行分支，此过程要保证结点不纯度最小，直至每个节点不纯度达到最小；

4) 生成多个决策树分类器，输入需要预测的元组，每个决策树都给出了一个分类结果，根据所有决策树的分类结果得出最终的类别。

3. 京沪深房地产上市公司盈利能力实证分析

3.1. 指标选取及数据获取

3.2. 盈利能力分级

Table 1. Quartile of the indicator

Table 2. Profitability score

${A}_{1}=\mathrm{min}+\frac{\mathrm{max}-\mathrm{min}}{3}*1=16.33$${A}_{2}=\mathrm{min}+\frac{\mathrm{max}-\mathrm{min}}{3}*2=25.67$

Table 3. Number of companies with various profit levels

Table 4. Number of companies with different profit levels in each region

3.3. 随机森林盈利能力分类模型

Table 5. The relationship between the number of classification indicators and the error

Figure 1. The relationship between the number of decision trees and the error

Table 6. Test results

3.4. 重要性分析

Table 7. Indicator importance measure

4. 政策建议

1) 坚持宏观调控方向

2) 加快实施房地产税收制度改革

3) 上市公司进驻三四线城市，开发有潜力的地产

Analysis of Profitability of Real Estate Listed Companies in Beijing, Shanghai and Shenzhen—Based on Random Forest[J]. 金融, 2019, 09(02): 96-103. https://doi.org/10.12677/FIN.2019.92011

1. 1. Ramanujam, V. and Venkatraman, N. (1987) Planning System Characteristics and Planning Effectiveness. Strategic Management Journal, 8, 453-468. https://doi.org/10.1002/smj.4250080505

2. 2. Kaplan, R.S. and Norton, D.P. (1992) The Balanced Score-card—Measures that Drive Performance. 187-204.

3. 3. Akhmedjonov, A. and Izgi, B.B. (2015) If Bank Capital Matters, Then How? The Effect of Bank Capital on Profitability of Turkish Banks during the Recent Financial Crisis. Macroeconomics & Finance in Emerging Market Economies, 8, 160-166. https://doi.org/10.1080/17520843.2014.940988

4. 4. 种镇国, 莫中杰. 基于因子分析的中国房地产上市公司绩效评价[J]. 经济研究导刊, 2010(36): 74-75.

5. 5. 王曼舒, 刘晓芳. 商业银行收入结构对盈利能力的影响研究——基于中国14家上市银行面板数据的分析[J]. 南开管理评论, 2013, 16(2): 143-149.

6. 6. 王吉恒, 王天舒. 基于企业层次分析法的盈利能力分析——以万科企业股份有限公司为例[J]. 哈尔滨商业大学学报(社会科学版), 2013(4): 3-10.

7. 7. Han Jiawei. 数据挖掘概念与技术[M]. 北京: 机械工业出版社, 2012: 249-250.

8. 8. 黄文, 王正林. 数据挖掘: R语言实战[M]. 北京: 中国工信出版集团, 2014.

9. 9. Liaw, A. and Wiener, M. (2002) Classification and Regression by Random Forest. R News, 23, 18-22. https://doi.org/10.1057/9780230509993

10. 10. Pavey, T.G. and Gilson, N.D. (2016) Field Evaluation of a Random Forest Activity Classifier for Wrist-Worn Accelerometer Data. Journal of Science & Medicine in Sport, 20, 75-80. https://doi.org/10.1016/j.jsams.2016.06.003

11. 11. 刘慧. 基于数据挖掘的电力上市公司盈利能力分析[D]: [硕士学位论文]. 北京: 华北电力大学, 2008: 35-36.