Selectively acquiring customer information: A new data acquisition problem and an active learning-based solution

This paper presents a new information acquisition problem motivated by business applications where customer data has to be acquired with a specific modeling objective in mind. In the last two decades, there has been substantial work in two different fields-optimal experimental design and machine learning-that has addressed the issue of acquiring data in a selective manner with a specific objective in mind. We show that the problem presented here is different from the classic model-based data acquisition problems considered thus far in the literature in both fields. Building on work in optimal experimental design and in machine learning, we develop a new active learning technique for the information acquisition problem presented in this paper. We demonstrate that the proposed method performs well based on results from applying this method across 20 Web usage and machine learning data sets.