Real world data is collected from multiple resources and there are high chances of having corrupt data. There might be missing values in the data set. Cleaning this data & filling up these voids is essential in order to build an efficient model.