Research On CBFI Prediction Based On Weighted Combination Model

The prediction research of Coastal Bulk Freight Index(CBFI)is not only a way for shipping enterprises to grasp the development situation of the bulk cargo transportation market,but also the basis of formulating enterprise development strategy.Therefore,the research on CBFI index prediction not only enables shipping enterprises to correctly grasp the market,but also enables the government to make reasonable planning and effective investment.It has great significance to grasp the chance and make decision for the future,meanwhile become the focus of the shipping market.CBFI is increasingly influenced by natural climate and market irregularities.The sequence of bulk freight is nonlinear and nonstationary.Experts and scholars on shipping industry are committed to exploring the high-precision and suitable prediction model in order to realize the sustainable development of the shipping market.CBFI released by Shanghai shipping exchange was taken as the research object.On the basis of the analysis of the internal fluctuation law of coastal bulk cargo transportation market,this thesis explored CBFI prediction model from the perspective of internal fluctuation characteristics.The main aim is to obtain a high-precision prediction model and provide technical support for the coastal bulk cargo transportation market.This thesis mainly discusses the following aspects:(1)The internal fluctuation characteristics of CBFI as the analysis object were analyzed using EMD method,IMFs reconstruction and statistical analysis method from three aspects:influence of significant events,the influence of irregular events on short-term market and the influence of long-term development trend.It establishes the foundation for the future research.(2)On the basis of CBFI sequence preprocessed by EMD,this thesis establishes the prediction models of PSO-LSSVM model and AR model.At the same time,compare with the single prediction model of LSSVM and PSO-LSSVM model,the prediction results show that the pre-processing combined model is more accurate and effective than the single prediction model.It is found that the predictive effect of PSO-LSSVM model preprocessed by EMD on the high value part is better than AR model and the predictive effect of AR model preprocessed by EMD on the stationary part is better than PSO-LSSVM model.(3)This thesis applies the above two methods which is AR model and PSO-LSSVM model preprocessed by EMD to establish weighted combination model in order to predict CBFI.The predictive results show that RMSE 20.3104,MAPE 1.3292,R2 98.14%and RE 1.33%,all of which are better than the combined prediction model based on EMD.Therefore,the weighted combination model can contain the important information in the prediction models and utilize comprehensively them to predict the future development trend of the coastal bulk cargo transportation market more accurately.After the case analysis,compared with the single model and the combination model without data preprocessing,the weighted combination model used in this thesis is better than the contrast model in terms of prediction accuracy and effectiveness,as to provide the technical model for accurately predicting bulk cargo freight rate.