Server load prediction based on improved support vector machines

To provide e-learning service more efficiently and effectively, Data mining technique have been applied in web-based distance education such as personalized service provision, server load prediction, etc. In web-based e-learning system, web server is the key and core component. In this paper, a novel server load prediction model is put forward by employing support vector machines (SVM). In addition, an approach to select free parameters of SVM is introduced which select parameters by checking if the training residual is white noise. Theoretical analysis and Experimental result has shown that by using this approach, server load prediction with high precision can be achieved.