摘要
工业增加值在经济中起着举重若轻的作用,其有效的现时预测有助于及时分析宏观经济走向。电力数据与经济生活密切相关,对经济预测有着重要作用。实时可得的电力数据使得工业增加值的现时预测成为可能。本文基于LSTM模型,研究如何有效利用电力大数据现时预测工业增加值。结果表明:LSTM与ARIMA模型结合、电力大数据和经济统计数据结合可得到最佳的预测效果;工业增加值现时预测中电力大数据与经济统计数据互补,两类数据结合的效果优于仅用某一类数据;电力大数据当期信息有助于提高预测效果;移动窗口整体表现比累积窗口好,这表明经济系统可能在不断变化。
Industrial added value plays a significant role in the economy,and its effective real-time forecast is beneficial to analyze the macroeconomic trend in time.Power data is closely related to economic life and plays an essential role in economic forecasting,of which real-time availability makes it possible to nowcast industrial added value.Based on LSTM model,this paper studies how to effectively use power big data in nowcasting industrial added value.Firstly,the best prediction comes from the combination of LSTM and ARIMA model as well as the combination of power big data and economic data.Secondly,power big data and economic statistics data are complementary,then combining two types of data is better than using single type of data.Thirdly,the current information of power big data is enough to improve the prediction.Fourthly,Moving Window performs better overall than Cumulative Window,suggesting that the economic system may be in flux.
出处
《价格理论与实践》
北大核心
2021年第7期110-114,共5页
Price:Theory & Practice
基金
国家电网有限公司大数据中心科技项目资助(SGSJ0000FXJS2000098)