There is trend to arrive at "single customer view" - combine all the available quantitative and qualitative data about single customer. How to organize those data for using in prediction and anylsis?

My guess is that all the data should be used for calibration of parameters of some model. But what models are available for representing single customer preferences and behavior? Does agent models from computer science (e.g. BDI - Beliefs-Desires-Intentions) can be used? Maybe neuroecnomics can be of help for building up such models (using choice, cost un intemporal discount functions)?