For mathematical decision making, analytic hierarchy process (AHP) has been regarded as useful. Extended AHP to be applicable to group decision making are called ”group AHP”. For the group decision making, it is important to aggregate opinions of all decision makers (DMs). Some of the group AHP methods determine the common weight of DMs and aggregate evaluated values with weight. In this paper, we propose a new group AHP method which uses own weight for each alternative, respectively, based on data envelopment analysis (DEA). However, more than one alternative may be judged as ”efficient” by basic DEA model. So that, we also extend our method once proposed for ranked voting model (Obata and Ishii, 2003) to the group AHP. By this method, we obtain the most favorable, normalized (i.e. has unit length) weight for each alternative.