A Gene Expression Programming Algorithm for Highway Construction Cost Prediction Problems

Abstract The gene expression programming as the new artificially intelligent and mathematic modelling method has very strong function discovery ability and high search efficiency. In view of characteristic of the highway construction cost forecasting and insufficiency of the existing forecasting model, highway construction cost forecasting model is constructed based on the GEP. Then, 11 highway engineering characteristic factors are determined according to the composition of building cost of projects. 15 groups of highway engineering data make the training sample and the confirmation sample. In the premise of configured the constitute elements of GEP algorithm, it is expressed by the Visual Basic program to solve the highway engineering construction cost forecasting model. The sensitive analysis demonstrates the model's feasibility, and the comparison chart and four target value evaluate the model. The result indicates the trend is consistent with the project practice, the predicted value and the actual value's relative error ≤5.9%, satisfies the forecasting precision (≤10%) and the forecasting precision is high, this model has the application value.