长租公寓装饰工程造价的快速预测研究

发布时间：2018-05-23 22:34

本文选题：长租公寓装饰工程 + 多元线性回归　；参考：《北京交通大学》2017年硕士论文

【摘要】：随着长租公寓的兴起,越来越多的企业投身到这一领域。目前,市场上多数长租公寓企业仍然靠着资本投资维持运转。但对长租公寓企业来说,如果不能实现正常的盈利,那么长租公寓企业将面对的是生死挑战,因为资本总是没有耐心的。长租公寓的盈利模式主要是房租差价和管理费,属于传统的重资产模式。因此,对成本的控制显得至关重要。装饰工程成本是整个项目成本的一大块,但是,传统的装饰工程估算过程复杂而繁琐,速度太慢,不能满足新形势下长租公寓企业的发展需求。本研究在装饰工程造价和机器学习相关文献的基础上,结合装饰工程造价管理工作的实习经验,并选取适用的机器学习模型实现了装饰工程造价的快速预测。本文首先界定了长租公寓装饰工程造价的内涵,明确了分析对象;然后通过系统的理论分析与实际工作经验相结合构建了装饰工程造价预测的指标体系;并选取了多元线性回归、BP神经网络和随机森林三种机器学习方法,系统分析了用这三种方法进行装饰工程造价预测的基本原理,构建了新的装饰工程造价预测模型;随后,选取了以往的装饰工程项目,用于装饰工程造价预测的数据集;最后将新模型通过数据集进行训练及预测。预测结果表明:随机森林模型的结果表现较好,预测精度较高,到达了比较好的预测效果,并且该模型具有较好的实用价值。本文将使用R语言作为数据分析和统计建模的工具,作为最近新兴起的一种计算机语言,它具备强大的图形处理和数据分析功能。
[Abstract]:With the rise of long-rent apartments, more and more enterprises devote themselves to this field. At present, most long-rent apartment companies in the market still rely on capital investment to maintain the operation. But for long-rented apartment companies, if they don't make a normal profit, they will face a life-and-death challenge because capital is always impatient. The profit model of long-rent apartment is mainly rent difference and management fee, which belongs to the traditional heavy asset model. Therefore, the cost control is very important. The cost of decoration project is a big part of the cost of the whole project. However, the traditional estimation process of decoration engineering is complicated and tedious, and the speed is too slow to meet the development needs of the long-term rental apartment enterprises under the new situation. On the basis of the related documents of decoration engineering cost and machine learning, this study combines the practical experience of decoration engineering cost management work, and selects the applicable machine learning model to realize the rapid prediction of decoration engineering cost. This paper defines the connotation of the cost of the decoration project of the long rental apartment, defines the object of analysis, and then constructs the index system of the cost prediction of the decoration project through the combination of the theoretical analysis of the system and the practical work experience. Three machine learning methods, multiple linear regression BP neural network and stochastic forest, are selected. The basic principles of cost prediction for decoration engineering are systematically analyzed, and a new cost prediction model of decoration engineering is constructed. The previous decorative engineering items are selected to predict the cost of decoration engineering. Finally, the new model is trained and predicted by the data set. The prediction results show that the results of the stochastic forest model are better, the prediction accuracy is higher, and the prediction effect is better, and the model has good practical value. In this paper, R language is used as a tool for data analysis and statistical modeling, and as a newly emerging computer language, it has powerful functions of graphic processing and data analysis.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU767

【参考文献】