Abstract: Support Vector Machine (SVM) is a novel learning method with solid theoretical basis depending on small amount of samples. Instead of the traditional inference process from induction to deduction, it greatly simplifies classification and regression problems. The decision function of SVMis only determined by a few support vectors, so the complexity of computation depends on the number of support vectors rather than the dimension of the sample space, thus avoiding the "curse of dimension" to some degree. Using SVM to process the fracturing data from a block in Daqing oilfield, a model for prediction of additional well production by fracturing stimulation is developed. The applied results indicate that SVM is a simple, feasible and effective method with high accuracy for prediction of incremental well production by stimulation treatments.

Key words: SVM, stimulation, stimulation treatments, incremental response, fracturing