Nowadays, choosing potential cities where to open new stores (especially supermarkets) and expand business plays a very important role for leading retailers due to the influence of huge and long-term investment. However, it is very time consuming for decision makers or experienced analysts to go through the whole city data and determine which city is better for a new supermarket. In the computer aided decision making scenario, retailers would like computers to model the data well enough to give a considerable rank result when some information is provided. In the real world new supermarket potential city selection problem, we find out that analysts can usually give some partially ordered information (i.e., for two cities, which city is better) and some relative comparison information (i.e., for three cities, the rank of city B is more close to city A than city C is to A) with high confidence. So in this paper, we propose a novel ranking model, which can deal with these two kinds of information together to evaluate the rank of cities. The promising experimental results in real world application data demonstrate the effectiveness of our methods for evaluating rank of cities for choosing new supermarket location cities. © 2011 IEEE.