Method for predicting bidding quotation of engineering project through feed-forward full-connection neural network

• Main Authors: XU YUNXIA, ZHAO YANHONG, PAN YANCHAO, LUO ZHI, LIN FEI, TAN ZHUO, HUANG XUETAO, CAI GUIJUN, ZHANG SEN, ZHANG KEFEI
• Format: Patent
• Language: Chinese; English
• Published: 24.08.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FORADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORYOR FORECASTING PURPOSES; PHYSICS; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE,COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTINGPURPOSES, NOT OTHERWISE PROVIDED FOR

The invention provides a method for predicting bidding quotation of an engineering project through a feed-forward full-connection neural network. The method comprises the following steps: dividing original data into training data and test data, and standardizing price-limited data in the training data; establishing a feedforward full-connection neural network regression model by taking the project limited price in the training data as input and the project quoted price in the training data as output; selecting k-fold cross validation to evaluate the model to obtain a k-fold cross validation result, and adjusting parameters of the model after each evaluation to re-evaluate; drawing an accuracy rate image, finding inflection points, performing early stopping, and the accuracy rate image is drawn with the round of model training as the abscissa and the k-fold cross validation result in each round as the ordinate; and testing the test data by using the adjusted model, and predicting the quoted price of the actual