Recent studies have demonstrated the effectiveness of proactive resource allocation under the assumption of perfect prediction of the user’s future data rate. In this paper, imperfect rate prediction merely based on the context information including large-scale channel gains of users and statistical information of system available resources is considered. Under the outagebased quality of service constraint, we propose a proactive power allocation policy aimed at minimizing the total transmit energy consumption. Simulation results show that the proposed statistical information-based proactive power allocation performs close to the one with perfect rate prediction, and exhibits evident performance gain over the non-proactive transmission policies.