Traditional elective system has structural deficiencies and defects. To avoid the fact that college students choose a course with blindness, therefore, with improved collaborative filtering algorithm, college students can get personalized elective course election. This paper first introduces two kinds of recommendation algorithms. Also the paper emphatically introduces recommendation algorithms based on collaborative filtering. It analyzes the advantages and disadvantages of the two algorithms. Finally, for data sparsity of collaborative filtering algorithm, it proposes an improved collaborative filtering algorithm, that adds factor in content-based collaborative filtering to solve this problem. Improved collaborative filtering algorithm avoids the traditional algorithms emerging data sparseness problem. Recommending appropriate courses for students on human-oriented, individual needs of students can be met.