Rough Set Processing Outliers in Cluster Analysis
Cui Guangcai, Hongwei Gao · 2019
Cluster analysis is a very important data mining technology, and also a hot issue in data mining research. Among many data types to be clustered, mixed attribute data is the most common one. Among them, the categorized attribute values are limited, disordered and can't be compared in size, which makes it difficult to determine the similarity measure that can reasonably describe the differences between sample objects. In addition, it is often impossible to convert class attributes into numerical attributes. At present, many numerical clustering algorithms are not suitable for processing mixed attribute data, but the number of algorithms that can process such data is small, and the performance and clustering quality still need to be improved.