Histogram-by: A grouping operator for continuous domains

摘要

In this paper, we propose a new operator, histogram-by, which provides a grouping for continuous domains, which partitions records into several groups by given ranges of the target attributes. The histogram-by operator can be represented as histogram-by clause in the SQL statement, and can be easily amenable to query optimization. As the application of the histogram-by operator, we introduce a multi-dimensional histogram query, which returns aggregate values of all ranges specified by the histogram-by clause. To process the query efficiently, we propose effective algorithms using aggregate R-trees. Our experimental results show that our algorithms are reliable in terms of performance over the synthetic and real-world datasets.