Carbon price fluctuation prediction using a novel hybrid statistics and machine learning approach

This study adopts a novel hybrid statistics and machine learning approach to predict carbon price fluctuations. We propose a framework integrating DILATED convolutional neural networks (CNN) and a long short-term memory (LSTM) neural network (NN) algorithm. We adopt the L2 parameter norm penalty as a regularization method based on statistics to make predictions based on the DILATED CNN-LSTM framework. Given the high correlation between the carbon price indicator and independent variables, we primarily include indicators related to blockchain information through the regularization process. We establish a dataset for carbon-price predictions. The experimental results indicate that the DILATED CNN-LSTM framework is superior to the traditional CNN-LSTM architecture and that blockchain information is associated with the carbon price. Compared to other approaches, the proposed RR-DILATED CNN-LSTM can effectively and accurately predict the fluctuation trend of carbon prices. The new forecasting methods and theoretical ecology proposed herein can provide a new basis for trend prediction and digital asset policy-making, represented by carbon prices, for both academia and practitioners.