基于决策树分类的可拓建筑策划预测方法  

Prediction methods for extension architecture programming based on decision tree classification

作　　者：郭强[1,2,3] 邹广天[1,2,3] 

机构地区：[1]哈尔滨工业大学建筑学院,黑龙江哈尔滨150006 [2]哈尔滨工业大学建筑计划与设计研究所,黑龙江哈尔滨150006 [3]黑龙江省寒地建筑科学重点实验室,黑龙江哈尔滨150006

出　　处：《智能系统学报》2017年第1期117-123,共7页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金项目(51178132)

摘　　要：为提升建筑师在策划过程中科学预测的能力,提出了一种基于决策树分类的可拓建筑策划预测方法。首先,运用数据采集软件批量采集互联网中的建筑案例数据,将数据预处理后存储至建筑案例库中;其次,通过评价特征选取、评价信息元集生成、决策树构建等步骤,获得决策树模型;最后,运用该模型预测当前策划项目的性能指标是否满足要求,并给出不满足要求情况下性能指标变换的途径。案例检验表明,该方法能有效提高建筑师运用互联网数据的能力,能够挖掘决策树分类知识,从而加速计算机辅助可拓建筑策划的进程。To improve the prediction ability of architects, apredic tion method for extension architecture programming （ EAP） based on decision tree classif icat ion was proposed. Fi r st,the architectural case data from the Internet were obtained by data acquisition software, and stored in anarch itec tura l case database after data preprocessing. Second,through evaluation characteristics selection, evaluation informat ion element set generation and decision tree construction, the decision tree model was discovered. Th e n, the performance indicators of the current project were predicted using this model,providing transformation approaches if the result did not satisfy the requirement. This study indicates that the proposed method can effectively improve an architects ability to use Internet data and mine decision tree classification knowledge,thus accelerating the process of computer aided EAP.

关 键 词：可拓建筑策划 决策树 分类 指标预测 可拓变换 

分 类 号：TU17[建筑科学—建筑理论]