一种基于改进时间卷积网络的生猪价格预测方法

针对传统的生猪价格预测方法存在预测精度不够高,容易陷入局部最小值等问题,为更加精准地预测生猪价格,采用随机森林回归(RFR)、极限梯度回升(XGBoost)、轻型梯度提升机(LightGBM)3种机器学习模型和改进网络结构的时间卷积网络(TCN)模型方法,以经过Z-Score标准化预处理的西南地区某省2011—2020年每周生猪价格数据为样本,对生猪价格预测进行研究。结果表明:TCN模型预测结果的均方误差(MSE)为0.340 606,平均绝对误差(MAE)为0.288 424,决定系数(R²)为0.995 683,均优于其他3种机器学习模型;与3种机器学习模型中效果最好的极限梯度回升(XGBoost)预测结果比较,3个指标分别提升了26%、8%和0.15%。改进网络结构的时间卷积网络模型可以更加精准地预测生猪价格。

A hog price prediction method based on improved temporal convolutional network

Insufficient prediction accuracy and easy to fall into local minimums are the problems existed in traditional hog price prediction methods. In order to predict hog price more accurately, three machine learning models, e. g. Random Forest Regression(RFR), Extreme Gradient Boosting(XGBoost)and Light Gradient Boosting Machine(LightGBM)were adopted in this study. The improved temporal convolutional network(TCN)model method was also used. The weekly hog price data from 2011 to 2020 in a province in Southwest China preprocessed by Z-Score standardization were taken as study object to investigate the hog price prediction. The TCN model prediction results were shown as follows: The mean square error(MSE)was 0. 340 606, the mean absolute error(MAE)was 0. 288 424, and the coefficient of determination(R²)was 0. 995 683, which were better than the other three machine learning models. Comparing the results of the improved temporal convolutional network model with the results of the three machine learning models, the three indicators of the most effective XGBoost were increased by 26%, 8%, and 0. 15%, respectively. In conclusion, the improved temporal convolutional network model can predict the price of hog more accurately.