Eliciting Attribute-Based Preferences From Scarce Big Datasets

The ability to estimate attribute-based preferences of individual customers is crucial for a wide variety of marketing tasks. However, the scarcity of Big Data arrays prohibits such estimation in many cases. In this paper, we present a novel method that allows eliciting individuals’ part-worths towards large number of product attributes from scarce big datasets by means of statistical techniques. Empirical tests on two real-world big data sets provide evidence that our estimates are reasonably accurate for predictions of future customer preferences.