摘要：基于特征价格理论，选择区位特征、建筑特征、邻里特征、消费者心理特征和租约特征五类共计20项特征变量，构建出长租公寓租金定价评估的随机森林模型，并将成都市长租公寓市场作为研究对象进行实证研究。研究结果表明：随机森林模型MAPE（%）为6.9%，优于神经网络和传统多元回归模型。
关键词：长租公寓；特征价格理论；随机森林；批量评估
Abstract：Based on the feature price theory，this paper selectsa total of 20 feature variables from five categories of location feature，building feature，neighborhood feature，consumer psychological feature，and lease feature to construct a random forest model for long-term rental apartment rental pricing evaluation，and takes Chengdu mayor’s rental apartment market as the research object for empirical research. The research results show that the random forest model MAPE（%）is 6.9%，which is better than neural network and traditional multiple regression model.
Keywords：long rent apartment；hedonic price theory；random forest；batch evaluation
建筑经济，2021（06）：99-102