基于非负矩阵分解的房地产评估模型

Assessment model of real estate appraisal based on non-negative matrix factorization

在房地产评估的基本方法——市场比较法中,所需统计经济指标较多且部分指标并不是消费者需求的普遍反映,该方法的使用与结果存在局限性.由于非负矩阵分解算法在对数据进行特征提取方面有较出色的表现,因此,文章首次将非负矩阵分解应用于房地产评估的数据处理中.根据分解结果,提取对房地产价值影响较大的几个指标因素进行归类,据此建立了新的市场比较法.经检验,其评估结果与实际市场成交价格接近,且新方法更简便易行.

The market comparison approach is a basic way of real estate appraisal.However,in this method,a lot of statistical indicators are needed,and some of them do not well reflect the consumers＇ demands.Therefore,this method has limitations.Considering the outstanding performance of non-negative matrix factorization algorithm in data feature extraction,the authors would like to apply it to data processing of real estate appraisal.According to the results of decomposition,several factors which influence the value of real estate to a large extent are extracted and classified,then a new market comparison method is established.The empirical analysis demonstrates that the assessment price of the new method is close to the transaction price,which indicates that the new method is more feasible.