Comparisons on Different Approaches to Assign Missing Attribute Values

A commonly-used and naive solution to process data with missing attribute values is to ignore the instances which contain missing attribute values. This method may neglect important information within the data, significant amount of data could be easily discarded, and the discovered knowledge may not contain significant rules. Some methods, such as assigning the most common values or assigning an average value to the missing attribute, may make good use of all the available data. However the assigned value may not come from the information which the data originally derived, thus noise is brought to the data. We introduce a new approach RSFit on processing data with missing attribute values based on rough sets theory. By matching attribute-value pairs among the same core or reduct of the original data set, the assigned value preserves the characteristics of the original data set. We compare our approach with “closest fit approach globally” and “closest fit approach in the same concept”. Experimental results on UCI data sets and a real geriatric care data set show our approach achieves comparable accuracy on assigning the missing values while significantly reduces the computation time.