Carbon Trading Price Forecasting with a Hybrid Arima and LSTM Deep Leaning Methodology

Abstract

China's carbon trading market is an emerging market and is influenced by many factors. Considering the regular characteristics of carbon trading price as an important indicator of market development, it is important to make scientific predictions about the price of carbon trading. Since more and more artificial intelligence models (including machine learning models and neural network models, etc.) have been applied to this field in recent years compared to previous studies that mainly used traditional econometric models. This paper tests and evaluates common artificial intelligence models and econometric models in the field of carbon trading price forecasting, taking the carbon trading price of Shenzhen from 2013 to 2020 as an example. It is found that the ARIMA -LSTM hybrid model has the best prediction effect, and the prediction results indicate that the Shenzhen carbon trading market will experience significant fluctuations in the next three years and the carbon trading price will remain unstable.