This paper introduces the decision tree algorithm into the field of preschool education to categorize the styles of children in preschool education. The learning activities of children with different styles are deeply analyzed by the total number of detections, the task score and the total game time in the small train by counting activity. Decision tree algorithm is utilized to integrate online preschool education resources and used in practice so as to assist teaching. The teaching experiment method is used to test its educational effect. Kindergarten children were categorized into 3 types: extroverted, negative emotional and effortful control children. Effort-control style children performed well in play detection behavior, play task score and total play time. In the teaching experiment, children in the experimental group obtained very significant improvements in small muscle activity, art, music and rhythm, blocks, natural science and mathematical thinking, while the control group also improved, but their changes were not significant. Decision tree algorithm has better results in assisting preschool education.