Research On The Prediction Of Project Cost Based On Principal Components Regression Model

With continuous deepening of market economic system, while the construction enterprises get more development opportunities, they should be confronted with more pressure of competition. Considering that the competitive advantage of the enterprises depends on noteworthy economic benefits, the scientific cost management which as a means of economic benefits is more important. To be the primary link of cost management, the cost prediction of projects can reflect the level of one construction enterprise’s cost management.With the tendency of project informatization, cost database of existing projects can provide a solid foundation for cost prediction of projects. The construction enterprises can find the potential knowledge which hides behind the large data, which will help them forecast the cost of each project effectively before they make investment decision. Based on the data of the sample projects, the author proposes to build a principal component regression prediction model.The main contents of the thesis are as follows:1. According to the large amount of cost database of existing projects and the composition of the project cost, the author selects the factors of cost prediction preliminarily in the angle of the qualitative and quantitative analysis. Furthermore, the author intends to build the prediction model by linear regression.2. In order to avoid the problem of multi-collinearity between multivariable factors which we have selected in step1by SPSS, the thesis comes to select the new public factors with the method of factor analysis. This method is a good way to reduce the indexes we need to analyze and build a simplified regression model, which will enhances maintainability and operation of the model.3. The author builds principal components regression model with the new indexes and tests the reliability of the model by the actual projects.4. At last, the thesis summarizes the results of the research and the disadvantages of the model. Moreover, it puts forward some prospects to improve the further related research.