Differentiating households to analyze consumption patterns: a data mining study on official household budget data

Abstract

Official data, administered by National Statistic Institutes (NSIs), play crucial role for being a major element of the governmental economic and social decision‐making process. This strategic role raises a significant necessity for statistical authorities to adopt new data tools to shift the statistical quality of the published data to a higher level. Data mining (DM) techniques and algorithms are promising tools to provide new ways to mine the crucial, complex, and voluminous official data to complement or substitute the traditional and lagged‐behind tools that NSIs have been using. This study addresses this potential utilization of DM tools on official data with a specific problem in an important survey for official statistics: Household Budget Survey. Through this study, clustering techniques are employed to characterize the household types and association rule mining technique is used to mine consumption patterns for each differentiated type. It is aimed to integrate the proposed model into data preprocessing procedure of the NSI to be able to engage in the real time analyses and to contribute exactness and timeliness of the data. WIREs Data Mining Knowl Discov 2018, 8:e1227. doi: 10.1002/widm.1227This article is categorized under: Algorithmic Development > Association Rules Application Areas > Government and Public Sector Application Areas > Industry Specific Applications