Potential of psychological information to support knowledge discovery in consumer debt analysisTools Ladas, Alexandros (2016) Potential of psychological information to support knowledge discovery in consumer debt analysis. PhD thesis, University of Nottingham.
AbstractIn this work, we develop a Data Mining framework to explore the multifaceted nature of consumer indebtedness. Data Mining with its numerous techniques and methods poses as a powerful toolbox to handle the sensitivity of these data and explore the psychological aspects of this social phenomenon. Thus, we begin with a series of transformations that deal with any inconsistencies the data may contain but more importantly they capture the essential psychological information hidden in the data and represent it in a new feature space as behavioural data. Then, we propose a novel consensus clustering framework to uncover patterns of consumer behaviour which draws upon the ability of cluster ensembles to reveal robust clusters from diffcult datasets. Our Homals Consensus, models successfully the relationships between different clusterings in the cluster ensemble and manages to uncover representative clusters that are more suitable for explaining the complex patterns of a socio-economic dataset. Finally under a supervised learning approach the behavioural aspects of consumer indebtedness are assessed.
Actions (Archive Staff Only)
|
Tools
Tools
