Abstract: Gerrymandering is the manipulating of redistricting for political gain. While many attempts to formalize and model gerrymandering have been made, the assumption of known voter preference, or perfect information, limits the applicability of these works to model real world scenarios. To more accurately reason about gerrymandering we investigate how to adapt existing models of the problem to work with imperfect information. In our work, we formalize a definition of the gerrymandering problem under probabilistic voter preferences, reason about its complexity compared to the deterministic version, and propose a greedy algorithm to approximate the problem in polynomial time under certain conditions.