With rapid economic growth,China has become the largest consumer of primary energy in the world.Energy possesses both commodity and financial attributes.To proactively address energy crises and financial risks,China is actively transforming its economic growth model and advocating a new paradigm of green development.The stock prices of the energy industry are the most direct and effective reflection of the game among stakeholders in the energy market,and fluctuations in energy prices have spillover effects,asymmetric effects,and aggregation effects.Taking China's Energy Index as the research object,this study innovatively introduces deep learning technology to organically integrate high-frequency and low-frequency data into a predictive big data set.By innovatively constructing the LASSO-LSTM-CNN deep learning hybrid model,the pre-diction accuracy has been significantly improved.The study results show that mid-to-long-term forecasts can modify the LASSO-LSTM or LASSO-LSTM-CNN to static multi-step output forecasts,which significantly outperform dynamic forecasts,with improvements in both accuracy and generalization capabilities.However,for long-term forecasts,as the explanatory power of high-frequency data gradually diminishes,it is necessary to comprehensively consider whether to use high-frequency data.China should recognize the fundamental and central role of energy in the industry chain from the perspective of ecology,and actively develop green and clean energy.At the same time,by fully utilizing the advantages of LASSO and LSTM-CNN mod-els,the accuracy of energy index prediction can be effectively improved,providing important references for financial decision-making.In mid-term forecasting,the positive impact of high-frequency data on forecasting ability should be fully considered,while in mid-to-long-term forecasting,the application of high-frequency data should be cautious.
关键词
LASSO-LSTM-CNN混合模型/能源指数/混频预测
Key words
LASSO-LSTM-CNN Hybrid Model/Energy Index/Mixed Frequency Prediction
维普
万方数据