基于熵权法-XGBoost的装配式建筑成本预测及控制研究

Research on the Cost Prediction and Control of Prefabricated Buildings Based on Entropy Weight Method XGBoost

以层数、建筑面积、结构类型、基础类型、预制率、构件种类、现场管理水平、人员技术水平和运距作为输入变量,以装配式建筑成本作为输出参数,使用决定系数(R2)、均方根误差(RMSE)和相对误差等评价指标对三种机器学习模型进行对比分析。结果表明:相比于BP神经网络模型和RF模型,XGBoost模型的决定系数更大,均方根误差最小,预测值与实际值的拟合程度达到0.9312,具有更好的预测性能。同时,引入熵权法对9种变量因素进行赋权分析,以此为基础,通过改变不同变量的取值范围,得到不同变量因素与成本的隐形关系,并以此实现多种情况下单方成本预测。

The number of floors,floor area,type of structure,type of foundation,prefabricated rate,type of components,level of site management,skill level of personnel,and transportation distance are used as input variables,and the unit cost of prefabricated buildings is used as an output parameter to build the model.Meanwhile,the three machine learning models are compared and analyzed using evaluation metrics such as coefficient of determination(R2),root mean square error(RMSE)and relative error.The results show that compared with the BP neural network model and the RF model,the XGBoost model has a better prediction performance with a fit of 0.9312 between the predicted and actual values.At the same time,the entropy weight method was introduced to analyze the assignment of nine variable factors.On this basis,by changing the value ranges of different variables,the invisible relationship between different variable factors and costs is obtained,and the cost prediction in many cases is realized.