摘要
缺失数据的填补是所有数据挖掘任务中非常关键的步骤。由于电力市场的复杂性,且电价的影响因素众多,造成电价数据在采集的过程中极容易产生有缺失值的现象,进而会影响到电价预测模型的建模效果。论文针对缺失电价数据,提出了一种马尔可夫链蒙特卡洛填补法(Markov Chain Monte Carlo,MCMC),并和常见的全条件定义法(Fully Conditional Specification,FCS)、MICE填补法在多种维度下进行了对比,实验证明MCMC填补方法在缺失电价数据填补上有一定的优势。
The filling of missing data is a very critical step in all data mining tasks.Due to the complexity of the electricity mar⁃ket and the influencing factors of electricity prices,it is very easy to generate missing values in the process of collecting electricity price data,which will affect the modeling effect of the electricity price forecasting model.In this paper,a Markov Chain Monte Car⁃lo(MCMC)method is proposed for the missing electricity price data,and the common Fully Conditional Specification(FCS)and MICE filling methods are used in various ways.The comparison is made under the dimension.The experiment proves that the MCMC filling method has certain advantages in filling the missing electricity price data.
作者
王曙
潘庭龙
WANG Shu;PAN Tinglong(Engineering Research Center of Internet of Things Technology Application,Ministry of Education,Jiangnan University,Wuxi 214122)
出处
《计算机与数字工程》
2020年第12期2954-2958,共5页
Computer & Digital Engineering
关键词
MCMC
缺失值填补
电价数据
数据挖掘
MCMC
missing value padding
electricity price data
data mining
