基于前馈神经网络预测猪肉价格的研究

Study on Predicting Pork Price Based on Feedforward Neural Network

本研究旨在通过分析2008年第1周至2023年第6周近15年全国猪肉价格波动规律,建立可预测未来一定时期猪肉价格变化趋势的模型,为政府和企事业单位进行市场分析提供参考。模型基于前馈神经网络(FNN),将类型变量处理成独热编码,数值类型变量采用标准化处理,并考虑价格延迟参数,经训练集训练后对测试集进行预测。结果表明,本研究所构建的模型能较准确预测猪肉价格,采用不同影响因素进行预测时结果有较大差异,仔猪价格是影响猪肉价格的关键因素之一,通过设置各影响因素的延迟参数可实现对历史数据的利用并扩大预测的时间跨度。

This study aims to establish a model that can predict the trend of pork price change in a certain period of time in the future by analyzing the fluctuation law of pork price in China i n the past 15 years from the first week of 2008 to the sixth wee k of 2023,so as to provide a reference for the government,enterprises and public institutions to conduct marketing studies.Based on feedforward neural network(FNN),a price forecasting model is constructed.In the model,the type variables are processed into unique thermal coding,the numerical type variables are treated with standardization,and the price delay parameter is taken into account.After training by the training set,the model has been used for predicting the test set.The results show that the model constructed in this study can predict pork price accurately,and the results are quite different when different influencing factors are used.Piglet price is one of the key factors affecting pork price,and the use of historical data and the time span of prediction can be expanded by setting the delay parameters of each influencing factor.