The missing data is an ever-present challenge faced by machine learning researchers while working on real-world datasets. Many such examples can be found; the UCI Machine Learning Repository hosts many datasets with missing values (Dua & Karra Taniskidou, 2017). Honeywell, (a well-known company that manufactures and services complex equipments) despite imposing regulatory conditions for data collection, had an industrial database which contained around 50% missing data (Lakshminarayan et al., 1999). The problem is more prominent in medical datasets related to patients’ health records, and in most of the cases the data is collected in an unorganized manner resulting into considerable information loss (Cios & William Moore, 2002). Almost every entry in these databases can have important values missing. In the case of wireless sensor networks, due to sensor failures or power outage, incomplete data is unavoidable (Gruenwald et al., 2010).