Blended learning relies heavily on pedagogical behavior research, however the issue of incorrect behavior assessment persists. The research issue of mixed teaching behavior in blended education cannot be solved by classical genetic algorithms, and the research results are not adequate. Consequently, this work reviews the literature on blended teaching behavior and suggests future research that will use the K-means clustering technique to study it. In order to minimize interference variables in teaching behavior study, the indicators are first classified according to the needs of the field, and then the influencing elements are located using cluster analysis theory. After that, a research plan is developed based on cluster analysis theory to study the teaching behavior, and the research’s findings are carefully examined. In terms of accuracy and speed of influencing elements in teaching behavior study, the MATLAB simulation results demonstrate that the K-means clustering method outperforms the classic genetic algorithm under specific assessment criteria.