基于ARIMA-GRNN组合模型的汽车零部件需求预测研究

Research on Demand Forecast of Auto Parts Based on ARIMA-GRNN Combined Model

单一的预测方法难以准确预测市场需求趋势,通过构建ARIMA-GRNN组合需求预测模型提高预测精确度:首先利用ARIMA预测出每月需求数并计算出每月实际需求数与每月预测需求数的误差值,再利用GRNN神经网络对误差值进行函数逼近与拟合,将拟合值对ARIMA预测值进行修正后的结果即为最终预测值。性能评估显示组合模型可以较好帮助汽车零部件企业提高市场预测精度。

It is difficult to accurately predict the market demand trend by a single forecasting method. By constructing the ARIMA-GRNN combined demand forecasting model to improve the forecasting accuracy: firstly use ARIMA to predict the monthly demand and calculate the error between the monthly actual demand and the monthly forecast demand. The value is then used to approximate and fit the error value by the GRNN neural network. The corrected result of the ARIMA prediction value is the final predicted value. Performance evaluation shows that the combined model can better help auto parts companies improve market forecasting accuracy.