While the conventional DEA based production plans aim to minimize all the inputs consumption and maximize all the outputs production, there are many real world production systems may also generate undesirable by-products. One methodological difficulty associated with the previous DEA-based production planning models is how to incorporate undesirable factors in the planning models, while the simultaneous increase of desirable outputs and decrease of undesirable outputs could be considered. Based on the assumptions that demand changes can be forecasted for the next production season and inputs changes are under control, this paper propose a new DEA based model in a centralized decision making environment, which shares inputs/outputs changes among all of the units in such a way that individual efficiencies do not decrease. The proposed model ensures the improvement of individual units’ performances, in addition to the overall efficiency. An empirical example is used to illustrate the proposed model.