Some phenomena have two distinct aspects, “positive” and “negative”. Suppose that a dataset is available to describe these aspects. We propose a model to decide which one has the best evaluate value. First, based on the dataset, we offer a new partial approximation space with a special basic structure and lower and upper approximation operators. Then, assigning two intervals to the two aspects, we evaluate them with the possibility degree formula, an effective tool for Multi-Attribute Decision Making (MADM) methods under uncertain environments. An example is used to illustrate how the theoretical construction may work in practice.