基于集成学习的短期电价预测

Short-term electricity price forecast based on integrated learning

为进一步提高短期电价预测结果的准确性和可靠性,文章提出了一种运用Blending学习方式去集成不同基础学习器的短期电价预测模型,首先运用按比例切分的方式对数据进行分割和训练,重新建立一个Hold-out集合作为二级特征,然后使用新构建的特征和标签去训练新的基础学习器,进而得到最终的融合预测值。实验结果表明,相较于单一的回归模型,Blending集成模型具有更小的误差和良好的稳定性,为短期电价预测提供了新思路。

In order to further improve the accuracy and reliability of short-term electricity price forecast results,this paper proposes a short-term electricity price forecast model using Blending learning method to integrate different basic learners.Firstly,the data is segmented and trained according to scale,and a Hold-out set is re-established as a secondlevel feature.Then the newly constructed features and tags are used to train new basic learners,and the final fusion prediction value is obtained.The experimental results show that compared with the single regression model,the Blending integration model has smaller error and better stability,which provides a new way for short-term electricity price forecast.