基于特征融合的生鲜商品短期销量组合预测

Combination forecasting of short-term sales for fresh products based on feature fusion

生鲜产品由于保质期短、易腐易损等特点,对短期销量预测的准确度和可靠性要求极高.为此,本文综合时间、定价、竞价、新鲜度等多种微观层面因素,开展了特征工程分析,并在此基础上提出了生鲜商品销量的ARIMA-NARX组合预测模型.该组合模型首先利用ARIMA模型描述销量时间序列中的线性规律,然后借助衍生降维处理后的特征矩阵,采用NARX捕捉ARIMA残差中的非线性关系,并利用NARX残差预测结果修正ARIMA预测值.最后,将该组合模型的预测结果与ARIMA、NARX、ARIMA-NAR、SVM及回归决策树等模型预测结果及真实观测值进行对比分析,通过MSE/MAPE预测误差评价和DM检验,验证了该组合预测模型的预测能力合理性和有效性,并能较大幅度提高生鲜产品短期销量的预测精度.

Due to the short shelf life and perishable characteristics,short-term fresh products sales forecast requires high accuracy and reliability.This paper carries out a feature engineering analysis for fresh products involving time,pricing,competitive product pricing,freshness and other micro-level factors.On this basis,an ARIMA-NARX combination forecasting model of fresh product sales is proposed.This combination forecasting model takes advantage of the ARIMA to capture the linear rule in the sales time series,and adopts the NARX to describe the nonlinear relationship in the ARIMA residual with the feature matrix processed by feature creation and dimension reduction.Then,the NARX residual prediction result is used to correct the predicted sales of ARIMA.Finally,the prediction results of the combined model are compared with the real observation values and prediction results of ARIMA,NARX,ARIMA-NAR,SVM and RT models.The MSE/MAPE value and DM test verify the rationality and effectiveness of the ARIMA-NARX model,which could improve the prediction accuracy of short-term fresh product sales significantly.