Research On The Cost Prediction Of Residential Construction Based On Machine Learning

The rapid and accurate estimation of the construction cost of residential buildings has attracted more and more attention from researchers.The project cost prediction of residential building is of great significance for the feasibility study of the project.This paper compares and analyzes the methods of cost estimation at home and abroad,studies their rationality and shortcomings,and puts forward the research object,research method and research content of this paper.First of all,based on the analysis of relevant theories of residential construction cost,through the analysis of literature,the factors affecting residential construction cost are summarized.It is mainly divided into building characteristics,decoration,infrastructure,building materials and building characteristics,with a total of 53 indicators.Through the questionnaire survey of 9 experts and the setting of 5 screening conditions,16 influencing factors were finally determined.Secondly,the training data of residential construction cost prediction model is selected based on BP neural network.According to the time of residential construction,the height of the building,the structure type of the building,the basic type of the building,the seismic intensity,the number of floors of the building and other characteristics of the sequence of determination.On this basis,the model structure of BP neural network is finally determined.The model level of the residential construction cost prediction model is 3 layers,with 15 input parameters,1 output parameter,10 hidden layer nodes,and the learning rate of 0.1.Finally,15 input parameters were optimized by principal component analysis to form a PCA-BP neural network prediction model.40 groups of sample data were collected and preprocessed.30 groups were used as training samples and 10 groups as test samples.The calculation results show that the prediction accuracy of residential building cost of PCA-BP neural network is higher than that of BP neural network.The results show that the prediction model established by PCA-BP estimation method has high accuracy and can meetthe application requirements of project cost management and control,which has certain practical significance.