ALGORITMA K-NEAREST NEIGHBOR CLASSIFICATION SEBAGAI SISTEM PREDIKSI PREDIKAT PRESTASI MAHASISWA

Students college predicate is a form of achievement during the academic activity at college. This research is intended  to make predictions toward predicate students college achievement that will be acquired in the future. The process of predictions by using K-Nearest Neighbor Method (KNN). The attributes that are used in process predictions was gender, kind of stay, age, semester credit unit, and also grade point average. Therefore by applying Al-goritma KNN, the predictions based on the closeness from history of data training  to data testing can be done. To determined of this  attributes based on the result of previous researches that have similarities of case that validated by academic of Faculty Sains and Technology. The process of predictions toward students information system of 2014/2015 as a sample of data testing. The number of the data was 50. And based on the data of students information system of 2012/2013 as a sample of data training, the number of the data was 165 which produce the accuracy testing was 82%. The result of calculation algoritma KNN is implemented toward Early Morning System (EWS). The output of sytem built to serve as a guide for students to improve the achievement and predicate in the future.