Uplift Modeling from Separate Labels
This addresses a practical limitation in uplift modeling for applications like targeted marketing and personalized medicine, offering an incremental improvement over conventional methods.
The paper tackles the problem of uplift modeling when only one label per instance is available, proposing a novel method that works in this more practical setting and demonstrating its effectiveness through experiments with a mean squared error bound.
Uplift modeling is aimed at estimating the incremental impact of an action on an individual's behavior, which is useful in various application domains such as targeted marketing (advertisement campaigns) and personalized medicine (medical treatments). Conventional methods of uplift modeling require every instance to be jointly equipped with two types of labels: the taken action and its outcome. However, obtaining two labels for each instance at the same time is difficult or expensive in many real-world problems. In this paper, we propose a novel method of uplift modeling that is applicable to a more practical setting where only one type of labels is available for each instance. We show a mean squared error bound for the proposed estimator and demonstrate its effectiveness through experiments.