This paper studies the specific application of association rule analysis in teaching evaluation in colleges and universities and analyzes the existing cases of the Apriori algorithm applied to college management under association rules. Aiming at the performance bottleneck of the Apriori algorithm in dealing with complex databases, the AMC algorithm based on matrix compression is proposed, and the data storage form is optimized through transaction matrix mapping. The improved AMC algorithm is applied to correlation analysis of English teaching evaluation data in colleges and universities, focusing on the correlation relationship between teaching characteristics, teaching-related factors, and evaluation grades based on confidence level. A confidence level of 69% will be achieved when the teaching effect is good, and the evaluation grade reaches basic satisfaction. A confidence level of 71% can be achieved when the teaching management is excellent and the evaluation grade is satisfactory.