Clustering Labeled Data Based on the Granularity

Abstract

Nowadays, the amount of data is rapidly increasing, and various analyses are being conducted to predict consumer behavior. Clustering exists as a means of finding useful information from the users' data. Clustering is often based on distances between data, but most studies do not regard the size of concepts. This study proposes a clustering method that takes into account the size of concepts (granularity). Clustering data with different granularity enables us to find people's thinking, such as paradigm shifts.