| Since the Ministry of Housing and Urban-Rural Development issued the Notice on Printing and Distributing the Project Cost Reform Work Plan(Construction Office Standard(2020)No.38),the project cost reform in China has been gradually deepened.In the notice,it was emphasized that budget quota would be gradually stopped,and enterprises should determine the consumption of the quantity machine and the unit price of the manpower machine by themselves.The current pricing quota and the information price issued by relevant institutions are only for reference.Among them,the material information price is an important part of the project cost,the proportion is as high as 70%.Therefore,following the pace of reform,it is necessary and urgent to provide material information price preparation method reference and price prediction for the majority of construction engineering enterprises.On the basis of extensive literature reading and research,combined with the research status of the information price of engineering materials at home and abroad,this paper studies the collection method,calculation method and price trend prediction of the information price of engineering materials,systematically summarizes the collection channels and standards,and introduces the 3σ method,Grubbs method and Dixon method to screen and process the collected abnormal data.After the abnormal data were screened and eliminated,the entropy weight method was adopted to select the material supplier based on the information price of the same material from different suppliers from different origin,and then the material information price was determined.The combined prediction model was constructed by SARIMA and GM methods.In addition,on the basis of the former study,a platform for material information price is designed and implemented.PHP scripting language is used to embed the combination model and corresponding data into the information platform development,so as to provide users with real-time prediction function of information price.The completion of this paper mainly includes the following aspects:First of all,the paper investigates the collection channels of the material information price available in the market,and refines the selection principles and collection methods of the collection channels.Aiming at the common problems in data collation and collection,this paper introduces the relevant theoretical knowledge of statistics and provides three scientific data processing methods to screen abnormal data.Secondly,the price of the same material after screening out abnormal data may have multiple prices due to different factors such as origin and supplier.The coefficient of variation method in the objective weighting method is used to determine the corresponding weights of different prices.The coefficient of variation can reflect the dispersion degree of the price of a single material in a period of time.It is more scientific to assign weights based on the degree of dispersion over the mean of the data unit.Thirdly,Python technology is used to construct the time series prediction model which can carry out long-term prediction and the grey prediction model which can carry out short-term prediction respectively,and the reciprocal mean square error method is used to give the two prediction models the corresponding weight to build the combined prediction model,so as to meet the prediction accuracy of information price.Through repeated test and fitting verification with a large number of actual data,the MAPE value of the combined model increased by 5.28%and the RMSE value decreased by 12.64% compared with the single model.Finally,a Web platform of material information price based on B/S architecture is constructed,which is realized by using PHP and other related technologies.The price prediction module is added to the platform,and the prediction model is applied to the prediction module of the platform by combining PHP and Python languages,so as to realize the online prediction function of material information price. |
PDF Full Text Request