Predicting with a small amount of data: an application of fuzzy reasoning to regional disparities

Abstract

Purpose – The aim of the paper is to propose a methodology that allows researchers and practitioners to structure a small amount of data in a way which aids understandings and allows predictions to be made. Design/methodology/approach – The paper explores how formal concept analysis can be combined with fuzzy reasoning to make predictions based on small datasets. A dataset of nine regions in Norway described by six attributes is used. The paper focuses on regional disparities in labour market outcomes such as unemployment and wages. Findings – The paper finds that unemployment tends to be concentrated in the most prosperous parts of the study area. These regions have high incomes and experience population growth. More rural regions have virtually no unemployment. The methodology proposed allows these patterns to be seen. The authors made predictions with an accuracy rate of over 75 per cent. Practical implications – A common response to high unemployment in urban areas is to stimulate employment growth. The findings suggest that this will simply increase migration towards the cities. The net result will be no change in unemployment but an accelerated depopulation of more rural regions. Originality/value – To the authors' knowledge, this is the first application of fuzzy reasoning to the topic of regional disparities. The methodology aids in the interpretation of small datasets. The methodology should be of interested to practitioners at the local level, who are interested in analysing their own region, even when limited data are available.