Bayesian based inference of missing time series values using Genetic Algorithm

Abstract

Time series datasets often suffer from the problem of non-ignorable incompleteness which needs careful attention. Even though researchers introduced many methods to handle the missing values, extra effort is needed in searching suitable method for discrete and continuous attributes. This paper proposes a new mechanism called as Bayesian Genetic Algorithm (BAGEL) which is capable of handling missing values in both continuous and discrete attributes in time series datasets using Bayesian analysis and Genetic Algorithms. In BAGEL, Bayesian principles are used to model the data and Genetic Algorithm is used to search the most accurate value that can be estimated from the available data. The method is applied to impute the missing values at different missing rates and the results produced are found to be more accurate when compared with existing methods.