Carbon price prediction based on integration of GMDH, particle swarm optimization and least squares support vector machines

Aiming at the problems of determining the inputs and parameters for least squares support vector machines (LSSVM)modeling, this paper presents an integrated model of group method of data handling (GMDH), particle swarm optimization (PSO)and LSSVM, i. e., GMDH-PSO-LSSVM, for inter- national carbon price prediction. First, GMDH is used to make the selection of input-layer units easily. Next, PSO is used to train LSSVM model with the training samples and obtain the optimal parameters. Then, the trained LSSVM is used to forecast carbon price of the testing samples. Finally, taking two carbon futures prices with different maturity called DEC 10 and DEC 12 of European Union emissions trading scheme (EU ETS) as samples, empirical results show that the proposed model is an effective way to improve forecasting accuracy.