To find the valuable information hidden behind the teaching evaluation data, a scheme of applying the data mining technology based on association rules to teaching evaluation is proposed. We analyze the basic principle of association data mining and the needs of teaching quality evaluation in colleges, and establish a multidimensional comprehensive evaluation online analysis mining model. After data standardization, to reduce the generation of candidate item sets, Apriori algorithm is optimized, and the running speed of association rule mining is increased by an order of magnitude. Finally, the concrete construction methods of mining environment and association rules are provided through experimental analysis, which verifies the effectiveness of the teaching evaluation scheme proposed in this paper.