That may be a part of it. The problem was only described in vague terms, so there's probably a reason why a Kalman filter makes sense, but as it was described in the interview it seemed more like a regression problem to me.