Abstract: Logistics prosperity index is a leading indicator reflecting economic development. Accurate prediction of logistics prosperity index is of great significance for assisting government to scientifically formulate economic regulation policies and guiding enterprises to conduct operational activities. A combination forecast model based on PSO-SVM (particle swarm optimization-support vector machine) is proposed, and the training set and the test set of the single forecast model are adjusted dynamically. The average value of the two adjacent single models is calculated as test (prediction) value of the total model. Taking logistics prosperity index forecast of Fujian province as an empirical subject, the relative error of the root mean square in the modeling stage is 1.26%, and the relative error of the root mean square in the testing stage is 0.82%. The results show that the accuracy of fitting and testing of the combination forecast model based on PSO-SVM is high.

Key words: logistics prosperity index, combination forecast, support vector machine, particle swarm optimization