Maintaining sttudent graduationn rates are the mmain tasks of a University. Highh rates of studennt graduation andd the quality of graduatetes is a success inddicator of a univversity, which will have an impactt on public confiddence as stakehollders of higher edducation and the Nationnal Accreditationn Board as a reguulator (governme nt). Making preedictions of studeent graduation a nd determine thee factors that hinders willl be a valuable input for Univeersity. Data minning system facililitates the University to create thhe segmentation of students’ performance and prediction of their graduaation. Segmentatition of student byby their performaance can be classsified in a quadrrant chart is divided intto 4 segments b ased on grade ppoint average annd the growth rrate of students performance in dex per semesteer. Standard methodologyy in data miningg i.e CRISP-DMM (Cross Industryry Standard Proccedure for Data MMining) will be iimplemented in tthis research. Making prredictions, graduaation can be donee through the moodeling process byy utilizing the coollege database. SSome algorithms such as C5, C & R Trree, CHAID, annd Logistic Regreession tested in o rder to find the bbest model. This research utilizes student performaance data for several classses. Parameters uused in addition to GPA also inccluded the masterr's students data are expected to bbuild the studentt profile data. The outcomme of the study iss the student categegory based on thheir study performmance and predicction of graduatiion. Based on thiis prediction, the universsity may recommeend actions to be taken to improvve the student achievement index and graduation rates.