基于深度学习的乘用车市场预警模型研究
Research on Chinese Automobile Market Warning System Based on Deep Learning
经过半个多世纪的发展, 中国已经成为全球最大的汽车生产基地以及消费市场. 在此背景下, 如何对汽车市场销量进行预警备受关注, 但是至今尚无精确有效的预警模型. 文章基于分位数广义自回归条件异方差模型(QGARCH)与长短期记忆网络模型(LSTM)构建了一个针对国内乘用车市场的精确有效的预警系统. 第 一步, 文章根据QGARCH模型构建了具有时变性与较强解释性的预警区域. 第 二步, 通过对已有的数据进行处理后, 根据LSTM模型对国内乘用车市场销量的同比增速进行预测. 文章将LSTM预测结果与支持向量机(SVM)、随机森林(RF)、极端梯度提升树(Xgboost)等模型的预测结果进行了比较, 验证了LSTM模型在乘 用车市场销量预测上的优势. 最后, 基于QGARCH模型与LSTM模型构建了一个精确有 效的预警体系. 结果显示, 文章构建的预警体系能够准确预测未来乘用车市场的预警情况且具有较强解释性, 对汽车企业具有极大的参考价值.
After developing for more than half a century, China has become the largest automobile production base and consumer market in the world. The topic about how to warn the underlying risk in automobile industry has attracted much attention, but there is no effective warning model. Based on quantile generalized autoregressive conditional heteroscedasticity model (QGARCH) and long and short term memory network model (LSTM), we build an accurate warning system for Chinese automobile market. Firstly, we build time-varying and warning zones via QGARCH model. Secondly, we predict the sales volume growth via LSTM and compare the results with other machine learning models, like support vector machine, random forest, extreme gradient boosting tree, by mean square error of prediction. Finally, we propose an effective warning system based QGARCH model and LSTM model. The results confirm that the proposed warning system can significantly improve the warning accuracy and provide valuable suggestion for enterprises.
汽车市场 / 分位数广义自回归条件异方差模型 / 深度学习 / 长短期记忆网络模型 / 预警体系. {{custom_keyword}} /
/
〈 | 〉 |