University administrators need graduation predictions to determine initial steps to avoid dropout. The length of a student's study period is determined by various factors. Therefore, it is important to know which students may not graduate on time. Data mining techniques can be used to explore new insights to predict student graduation. By using the association rule technique we can obtain information from large data such as data from universities. The aim of this research is to determine the pattern of study duration for Graha Nusantara University F-KIP students. by using the association rule data mining method and comparing a priori algorithms and hash-based algorithms. The data used is Graha Nusantara University F-KIP master data which is processed using association rule data mining techniques with a priori algorithms and hash-based algorithms with minimum support of 1% and minimum confidence of 1%. The results of data processing with the a priori algorithm are the same as the results of data processing with the hash-based algorithm, namely 49 2-itemset combinations. The pattern that was formed included 7.5% of graduates from the mathematics department studying for more than 5 years with a confidence value of 38.5%.