Conventional testing systems usually give students a score as their test result, but do not show them how to improve their learning performance. Researchers have indicated that students would benefit more if individual learning guidance could be provided. However, most of the existing learning diagnosis models ignore the fact that one concept might contain multiple knowledge levels with different degrees of difficulty, and hence students might be guided in an inefficient and ineffective way. In order to provide more precise learning guidance to individual students, the study described in this paper uses an enhanced concept-effect model for diagnosing students' learning problems and providing learning advice. The experimental results from a mathematics course have demonstrated the utility and effectiveness of this innovative approach. (Contains 1 note, 2 tables, and 5 figures.)