IMPLEMENTASI LOGIKA FUZZY MAMDANI DALAM MEMPREDIKSI PELUANG KELULUSAN CALON MAHASISWA BARU JALUR SELEKSI NASIONAL BERBASIS PRESTASI (SNBP)

Abstract

Uncertainty in estimating the graduation probability of Seleksi Nasional Berdasarkan Prestasi (SNBP) often triggers ambiguity for prospective new students in determining their study program selection strategy. This is caused by the non-linear and highly variable nature of the graduation parameters. This study aims to implement the Mamdani method's Fuzzy Inference System (FIS) to predict the percentage of SNBP graduation opportunities quantitatively. The fuzzy model is designed using three crisp Input s: the average report card score over 5 semesters, the number of applicants in a specific study program, and the intake quota. The Output variable is the probability of passing, scaled from 0 to 100 percent. Data processing involves fuzzification using linear membership functions (triangular and trapezoidal), constructing a rule base comprising 64 conditional IF-THEN rules, an inference process using max-min composition, and a defuzzification stage utilizing the Centroid method (Center of Area). Model validation was tested using a case study with Input  parameters: an average 5-semester score of 88 (Normal and High categories), 300 applicants (Moderate and High categories), and an intake quota of 25 seats (Low and Moderate categories). The test results produced a post-defuzzification crisp Output value of 54.9%. Based on these results, the system successfully and objectively predicted the acceptance probability of the test subject, classifying it into the "Moderate" category. This study proves that the Mamdani fuzzy logic method is adaptive and reliable as a transparent decision-support instrument for prospective college applicants.