基于EMD-ARIMA组合模型的长江航运干散货运价指数预测

Prediction of the Yangtze River Shipping Dry Bulk Freight Price IndexBased on the EMD-ARIMA Combined Model

针对长江航运干散货运价指数(YBFI)样本点少、周期性不明显、非线性,以及非平稳特性,从分析数据内在波动性出发,提出一种基于经验模态分解(EMD)-差分整合移动平均自回归(ARIMA)组合模型.对比传统ARIMA模型、简单季节预测两种方法,EMD可对YBFI序列进行降噪分解,保留数据的内在特性;分解后的序列用ARIMA模型、三角函数拟合,效果良好.重组后分析误差,发现该组合预测模型与传统单一模型相比误差较小,预测精度更高.

According to the characteristics of Yangtze River Shipping Dry Bulk Freight Index(YBFI), such as few sample points, unobvious periodicity, nonlinearity and non-stationarity, a combined model based on empirical mode decomposition(EMD) and differential integration moving average autoregressive(ARIMA) was proposed from the perspective of analyzing the inherent volatility of data. Compared with traditional ARIMA model and simple seasonal prediction, EMD can denoise and decompose YBFI series, and keep the intrinsic characteristics of data. The decomposed sequence is fitted by ARIMA model and trigonometric function, which has a good effect. After the error analysis, it is found that the combined forecasting model has smaller error and higher forecasting accuracy than the traditional single model.