Prediction Model of the Development Trend of Chinese Development Finance Based on CEEMD–LSSVM

Development finance can achieve the development goals of the government in the market through the combination of financing and government organization advantages. The prediction of development finance development trend is of great significance for the national and local governments to make relevant decisions. In this work, the core data of development finance, the balance of medium and long-term loans, are selected as the research object, which are nonlinear and non-stationary and have a small sample size, making them difficult to accurately predict. A combination model based on complete ensemble empirical mode decomposition (CEEMD)–least squares support vector machine (LSSVM) is proposed to analyze the development trend of development finance in China for improving the prediction accuracy and effectively explaining development finance loans fluctuations. First, the loan balance could be decomposed into several intrinsic mode component, which solves the nonlinear and non-stationary problems, based on CEEMD. On this basis, LSSVM was used to separately forecast each component. Finally, the development trend of developmental finance was obtained by reorganizing the forecast results of each component. The prediction results of the CEEMD–LSSVM model are compared with support vector machine, BP, and so on, which show that the proposed model has higher accuracy, to verify the effectiveness of the model.