Cost Impact Factors and Control Measures of Road and Bridge Projects Based on Linear Regression Model

In the process of road and bridge project cost prediction, factors such as material, labor, management and design can affect the prediction results. In order to accurately understand various influencing factors on the projects’ cost, a model using principal component regression and support vector machine is established. It uses principal component analysis to find the factors that have the greatest influence on the project cost prediction and to predict the cost of each factor to determine the magnitude of the influence of each major factor; meanwhile, the study compares the principal component regression-support vector machine model with the principal component regression model and the support vector machine-radial basis neural network model. The results of the study show that three models start to converge after about 120, 100 and 70 iterations, respectively, when the loss values are about 0.3, 0.25 and 0.16, respectively; the lowest R-squared of the three models for the five costs are 0.78, 0.75 and 0.87, respectively. It can be seen that the convergence and fit of principal component regression-support vector machine are better than those of the other two. Meanwhile, among the total costs, infrastructure costs, material costs, labor costs, equipment costs and management costs accounted for 5%, 63%, 13%, 12% and 7%, respectively; the material factor had the greatest impact on the total costs.