Customized SQL Aggregations for Generating Data Sets

Abstract: Data mining is the domain which has utility in real world applications. Data sets are prepared from regular transactional databases for the purpose of data mining. However, preparing datasets manually is time consuming and tedious in nature as it involves aggregations, sub queries and joins. Moreover the traditional SQL aggregations such as MAX, MIN etc. can generate single row output which is not useful in generating datasets. Therefore it is essential to build horizontal aggregations that can generate datasets in horizontal layout. These data sets can be used further for data mining in the real world applications. This paper focuses on building user-defined horizontal aggregations such as PIVOT, SPJ and CASE whose underlying logic uses SQL queries. We built a prototype that demonstrates the efficiency of the proposed horizontal aggregations. The empirical results revealed that the prototype is effective and can be used in real world applications.

Keywords: SQL, aggregations, horizontal aggregations