基于大数据的碳价预测

为了研究大数据是否能够帮助我们预测碳排放权价格，本文讨论了结构化数据和非结构化信息对预测碳价所起的作用。结构化数据选取了国际碳现货价格、碳期货价格和汇率，非结构化信息选择百度搜索指数和媒体指数。考虑到当解释变量很多时，平等对待每一个解释变量是不合理的，所以提出了网络结构自回归分布滞后（ADL）模型,在参数估计和变量选择的同时兼顾了解释变量之间的网络关系。实证分析表明，网络结构ADL模型明显优于其他模型，可以获得较高的预测准确性，更适合基于大数据的预测。

Forecasting of Carbon Price Based on Big Data

This paper analyzes the effect of structured data and unstructured information on carbon price forecasting in order to learn if big data can help us predict carbon price. We choose international carbon spot price, carbon futures price and exchange rate as structured data, Baidu search index and media index as unstructured information. Considering that it is not reasonable when there are a lot of explanatory variables by taking every variables equally, we proposes network autoregressive distributed lag(ADL) model, taking variables as a network when we estimate parameters and choose variables. The empirical results show that network ADL model is better than other models, which can get high accuracy, and it is more suitable for predicting based on big data.