Generic Framework for Gaining Insight Into Data

Abstract

Efficient data analysis can be made easier with datasets having columns in horizontal tabular layout. Aggregations using standard SQL return one column per aggregated group. So existing SQL aggregations have limitations in preparing datasets. In this paper we have proposed a framework to build dataset using a new class of functions called horizontal aggregations. To speed up the dataset preparation task we have partitioned vertical aggregations on grouping column and optimized SPJ method. Also it is proposed to integrate summary dataset, obtained from the result of horizontal aggregation, into homogeneous cluster using K-means algorithm.
DOI: 10.17762/ijritcc2321-8169.1506164