摘要
时间序列分析是在气象预报、地质变化分析、交通流量预测和财经市场预测等应用中的重要分析手段之一,但是在时间序列的数据中经常会发生观测数据缺失的情况,例如传感器故障造成的缺失,地质数据中部分相关年份的观测数据缺失等,目前如何使用相应的方法和模型对缺失的数据进行填补和使用含有缺失数据的时间序列进行预测是目前该方向研究的一大热点。基于此,笔者将列举一系列目前所使用的缺失数据的处理方法及其优缺点。
Time series analysis is one of the important analysis methods in the application of weather forecast,geological change analysis,traffic flow forecast and financial market forecast.However,in the use of time series data,the lack of observation data often occurs,such as the loss caused by sensor failure,the lack of observation data in some relevant years in geological data,etc.How to At present,it is a hot topic to fill the missing data with corresponding methods and models and to predict the missing data with time series.The author will list a series of processing methods of missing data and their advantages and disadvantages.
作者
陈雁声
Chen Yansheng(Shanghai Hongkou District Amateur University,Shanghai 200080,China)
出处
《信息与电脑》
2020年第10期19-22,共4页
Information & Computer
关键词
时间序列
缺失数据
预测
数据填补
time series
missing data
prediction
data filling
