Measuring wellbeing by extracting social indicators from big data

The construction of social indicators is based upon the availability of data collected on purpose (official statistics); it is a common view that constructing social indicators, in the perspective of measuring wellbeing of countries, can benefit from the availability of new sources of data (e.big data). One of the big challenges in dealing with new data sources is related to possibility to describe complex human and social phenomena from different perspectives; this introduces new issues to be faced in constructing indicators. We explore how the classical methodological approach to social indicators construction should be re-considered in light of using data collected not on the purpose of constructing indicators but for other aims (e.g., administrative). In this sense, the individual sales receipts, collected during the period 2007-2014 and made available to our group by a big Italian chain of stores (food but not only), allow us to explore not only a particular social phenomenon but also the methodological implications in dealing with big data. We extend our previous analysis done by using data mining techniques (e.g., clustering analysis) showing some buying behavior in the period of crisis that suggested us some insights regarding the various categories of products that customers purchased, in order to identify some typical purchase behavior but also to test if and how this information allow other information to be estimated (e.g., structure of the household).