基于组合模型的服装定制面辅料预测方法

Prediction Method of Garment Customized Surface Accessories Based on Combination Model

针对服装定制企业中根据工作人员个人经验采购面辅料造成面辅料过剩与不足的问题,以及仓库库存资源被过剩面辅料长期占用所导致仓库利用率过低的问题,构建了基于GS-ARIMA-GARCH的服装定制面辅料需求量预测模型,将网格搜索、ARIMA模型和GARCH模型相互组合在一起,进一步提升服装定制面辅料预测模型的预测精度。实验结果表明,引入GARCH模型可以很好地消除面辅料残差序列中出现的异方差现象;通过对GS-ARIMA预测模型和GS-ARIMA-GARCH预测模型进行精确性对比分析,利用评价指标RMSE值和MAE值对其进行判断,从结果可以看出GS-ARIMA-GARCH模型的预测准确精度相对于GS-ARIMA模型更加精准,面辅料需求量预测效果更好;通过对实际面辅料时间序列进行预测分析,从结果看出预测值与实际值的相对误差值在0.5%~2.5%范围内并且R2值结果为0.905 504,可以准确地预测出短期内面辅料的需求量,为企业制定合理的面辅料采购计划,从而提升仓库库存资源的利用率。

Aiming at the problem of surplus and shortage of surface materials caused by purchasing surface materials based on personal experience of staff in garment customization enterprises, and the problem of low utilization rate of warehouse caused by long-term occupation of warehouse inventory resources by surplus surface materials, we construct the demand prediction model of garment customization surface materials based on GS-ARIMA-GARCH. Grid search, ARIMA model and GARCH model are combined together to further improve the prediction accuracy of clothing customization surface and accessories prediction model. The experimental results show that GARCH model can eliminate heteroscedasticity phenomenon in residue sequence of surface and auxiliary materials. The accuracy of GS-ARIMA prediction model and GS-ARIMA-GARCH prediction model was compared and analyzed, and the evaluation indexes RMSE value and MAE value were used to judge them. It can be seen from the results that the accuracy of GS-ARIMA-GARCH model is more accurate than that of GS-ARIMA model. The demand forecasting effect of flour and auxiliary materials is better. Through analyzing actual surface materials time sequence forecast, seen from the results and the actual and estimated values of the relative error value is within 0.5%~2.5% and R2 value is the result of 0.905 504. It can accurately predict the short-term demand for surface materials, which formulate reasonable surface materials purchasing plan for the enterprise, so as to further enhance the utilization of the warehouse inventory resources.