基于QAR-EGARCH模型组合预测的气温研究——以合肥市为例

Combined Forecasting of Air Temperature Based on the QAR-EGARCH Model——A Case Study of Hefei

在天气预报中,常存在着一定的误差,其波动也出现着一定的波动集聚性和不对称性。因此有学者基于气温波动特征建立了AR-EGARCH模型,较好地捕捉了气温波动趋势,获得了较好的效果。但是它是均值意义下的回归模型,其稳健性较弱。本文将分位数回归和它进行结合,并针对每个分位数点的预测进行加权组合。结果表明,组合模型的预测精度有一定的提高。

There often exist some errors in weather forecast,and the volatility of temperature also shows clustering and asymmetry.Therefore,the AR-EGARCH model based on the feature of temperature volatility was proposed to better capture its trend and it obtained better results.However,the AR-EGARCH model is a regression model in terms of the mean with weak robustness.So it is combined with the quantile regression,and the forecast result is weighted and combined in each quantiles.The result shows that the model forecast accuracy is improved after the combination.