Accurate financial aid is an important content and due meaning of college students’ financial aid in colleges and universities in the new development stage. The article uses the campus card of University W as the research data source and uses the FSIPD algorithm to search and process the relevant features of poor students after modeling the poverty data. In the initial feature set, the extraction of students’ average daily consumption amount is realized by entropy and proximity, and the MD-KNN algorithm is combined to classify the average daily consumption amount of poor students. The strong association rules of poor students’ data were mined by the association rule Apriori algorithm, and the differences between the college students’ financial aid policies and the categories of the funded groups were investigated by combining the FCM cluster analysis algorithm. The strong association rules of “especially poor”, “poor” and “generally poor” students are obtained by the Apriori association algorithm, and their support and confidence levels are kept high. The level of support and confidence for these rules is high. The learning status of the subsidized group is divided into three clusters, namely, stray, active and lazy, and there is a significant difference (Sig<0.05) between the satisfaction of national scholarships and national student loans among stray, active and lazy students. Classification based on the MD-KNN algorithm yields that the average daily consumption of poor students resides between 20 and 27 yuan. Colleges and universities have the ability to create appropriate financial aid policies for the subsidized groups based on the data mining results to ensure the financial support for college students’ studies and lives.