标题:
基于增强回归树的房价影响因素分析—以波士顿地区为例Factor Analysis of Housing Price Based on Boosting Regression Tree—Taking Boston as an Example
作者:
盛佳, 潘东东
关键字:
回归树, 增强法, 房价, 因素分析Regression Tree, Boosting, Housing Price, Factor Analysis
期刊名称:
《Statistics and Application》, Vol.5 No.3, 2016-09-29
摘要:
房价是反映一个地区经济社会发展水平和状况的重要指标，对其影响因素以及影响的方式和程度进行探究具有理论价值和现实意义。增强回归树是近年来机器学习领域备受关注和推崇的一种非参数建模分析方法，具有建模效率高、模型结果易于解读等优势。本文以美国波士顿地区的历史房价数据为例，采用增强回归树方法来探寻该地区房价的主要影响因素，并比较不同因素在回归树中的相对影响强度。本文得出的结论可为我国某些中心城市的房价调控政策提供参考。
Housing price is a very important index which can reflect the economic and social development level and situation of a certain region or city. It is of great theoretical value and practical meaning to study important factors influencing housing price as well as their influence patterns and magnitude. Boosting regression tree has been recently developed as one of the most prevalent nonparametric modeling methods in the fields of machine learning, which has desirable properties such as high efficiency as well as easy-interpretation. In this paper, we take the housing price data in Boston as an example and try to analyze factors determining housing price based on Boosting Regression Tree method. We identify some relatively significant factors by comparing their relative importance in the model and also investigate their influence patterns. Results in this paper could be reasonably extended to housing price researches of some Chinese first-tire cities.