The traditional outlier mining methods are affected by man-made factors and mined outliers can not be analyzed further. In this paper, an outlier mining algorithm based on characteristic attribute subspace is presented. Firstly, the concept of attribute entropy is introduced to calculate corresponding attribute abnormal degree, characteristic attribute subspace and attribute weight. Secondly, subspace outlier factor is computed, and then outliers are found. The method does not depend on beforehand parameters or thresholds and can explain the meaning of the outliers clearly. In the end, experimental results validate the feasibility and effectiveness of the algorithm by utilizing UCI and high-dimensional star spectrum data.