Nowadays, e-training has been widely used in differentindustries. In the era of big data, utilizing data effectively toexplore useful information for learners and administrators is amore concerned problem in the field of education. This paperpresents the design and implementation of an e-training decisionmakingsystem based on outlier mining technology, which is animportant research field in data mining. By discovering smallgroups of abnormal learning behaviors that are exceptional whencompared with the rest large amount of learning behaviors, thesystem could identify those behaviors, then give specificsuggestions to learners and assist administrators in makingdecisions.