On imputation methods in univariate time series

Abstract

Missing data become the first obstacle when designing predictive models, as most statistical methods are premised on complete data without missing values. Therefore, when it comes to analyze the data, the missing values should be replaced with rational values, to carry out an analysis based on a "complete" dataset; in statistics this approach of handling missing values is called Imputation. Time series data exist in nearly every scientific field, where data are measured, recorded and monitored, so it is understandable that missing values may occur. Hence, time series characteristics need to be taken into consideration, to develop an appropriate and efficient strategy when dealing with missing data. The main scope of this work, is to compare and quantify the performance of imputation algorithms in the context of univariate time series data, using the statistical software R. Three basic types of time series patterns are considered and the imputation techniques examined are compared by means of two error metrics, namely, MRSE and MAPE.