Study On The Prediction Of Construction Cost Based On PCR-SVR

The competition is very fierce in the construction industry under background of the whole good development of situation. If construction companies want to avoid being defeated in competition, it is important to strengthen the engineering cost management. The cost prediction is an important part and content of the engineering cost management. It is especially critical to ensure the continuous, rapid and healthy development of the construction industry. Considering the present situation of cost prediction, there are problems that the predictions are inaccurate and the forecasting techniques are unscientific, a method of combining PCA with SVR is suggested to predict construction cost.The reasonable cost prediction index system is established in this paper, based on the studies of summarizing the domestic and foreign researches and literature review and the characteristics of cost.The theory and method of PCA and SVR is introduced in details. SVR regression estimate algorithm and PCA are applied to established forecasting model. There are many factors affecting the cost of building. We use PCA to eliminate correlation,remove overlapping information in index, and extract principal component which can reduce complexity of samples and improve calculating speed of SVR. The PCA-SVR integrated method has advantages at the ways of the principal component analysis and SVR small sample and non-linear features; it not only increases the speed of system and raises the precision of the prediction but also facilitates cost prediction.The model is applied to predict cost combined with concrete example. The results show that predicted values and actual values show high degree to match, which validates the model effectiveness and feasibility. It is of positive significance to improving the level of engineering cost management.