基于非结构化数据的LLE-WOA-LSSVR碳价格组合预测模型

LLE-WOA-LSSVR combination forecasting model of carbon price based on unstructured data

在传统的组合预测模型中,利用的数据大多为结构化数据,然而在网络环境下,非结构化数据广泛存在,因此充分利用非结构化数据所提供的有效信息是预测中要解决的关键问题之一。针对上述问题,文章构建了基于非结构化数据的局部线性嵌入和鲸鱼优化算法的最小二乘支持向量回归(locally linear embedding-whale optimization algorithm-least squares support vector regression,LLE-WOA-LSSVR)碳价格组合预测模型,通过LLE算法对非结构化的高维数据进行降维处理,并利用LSSVR进行预测。考虑到LSSVR模型中参数的选取会对预测结果产生影响,引入WOA算法优化模型中的参数。碳价格预测的实例结果表明,LLE-WOA-LSSVR预测模型可行且有效。

In the traditional combination forecasting model,most of data used are structured ones.However,in the network environment,unstructured data exist widely.Therefore,how to make full use of the effective information provided by unstructured data is one of the key problems to be solved in forecasting.In the above background,this paper constructs the locally linear embedding-whale optimization algorithm-least squares support vector regression(LLE-WOA-LSSVR)combination forecasting model of carbon price based on unstructured data.This model is characterized by LLE algorithm for dimensionality reduction of unstructured high-dimensional data and LSSVR for forecasting.Considering that the selection of parameters in LSSVR model will affect the forecasting results,WOA is introduced to optimize the parameters in the model.The example of carbon price forecasting shows that the LLE-WOA-LSSVR forecasting model is feasible and effective.