Abstract
We present a framework for recommender systems (RS) to support exploratory data analysis (EDA) in analytical decision making. EDA helps the domain expert, often not a statistical expert, discover interesting relationships between variables and thus be motivated to explain the data. By capturing the behavior of expert analysts in EDA, RS could advise domain experts of “standard” analytical operations and suggest operations novel to the domain but consistent in analytical goals with requested operations. We enhance our framework with rules that encapsulate standard analytical practice and by incorporating user preferences. We present a scalable framework architecture, which we implemented in a prototype system, and discuss two use cases where the prototype was exercised, analyzing data from image analysis and analyzing eye tracking data.