随机缺失数据下的时间序列分析建模

Time Series Modeling for Missing Data at Random

在时间序列建模过程中,数据的缺失会极大地影响模型的准确性,因此对缺失数据的填补尤为重要.选取北京市空气质量指数(AQI)数据。将其随机缺失10%.分别利用EM算法和polyfit直线拟合的方法对缺失值插补,补全数据后建立ARMA模型并作预测分析.结果表明,利用polyfit函数插补法具有较好的结果.

It is well known that the accuracy of a time series model greatly depends on the collected data,and therefore it is very important to handle the data with missing values.In this paper,based on the data of Beijing Air Quality（AQI）,10%of it is missing at random,the ARMA models are estabhshed by imputation method of missing data based on both EM algorithm and polyfit line-fitting algorithm.The results show that the polyfit algorithm is more effective.