Feature selection (FS) is an important task in data analytics. Feature selection is also referred as Attribute Reduction (AR) is a process of finding subset of features which are most predictive of a given result. In general, mining useful prediction from datasets that contain huge numbers of features as in gene database, protein structures, weather forecast etc. is most challenging task. Though several techniques for attribute reduction are in existence still there is a quest for novel approaches. Feature Selection techniques has tradeoff between the computational complexity and accuracy hence it is a challenging task. This paper presents analysis on Feature selection techniques based on Rough Set theory (RST) for attribute reduction preserving data originality. © Research India Publications.