Profilling model of students' mathematics learning performance using fuzzy inference system based on assessment results and adversity quotient

Abstract

This study develops a learning performance profiling model for mathematics students using the Mamdani Fuzzy Inference System (FIS) by integrating assessment results and Adversity Quotient (AQ). The study employs a quantitative approach with computational modeling design. Input variables consist of academic assessment scores (range 0–100) and AQ scores measured through the CORE dimensions (Control, Ownership, Reach, Endurance), while the output variable is the student learning performance profile categorized as Low, Sufficient, Good, and Very Good. Data were obtained from 31 students of SMAN 2 Telukjambe Timur. The FIS model was built through fuzzification, fuzzy inference using MIN-MAX operators, and centroid defuzzification, implemented using Python Scikit-Fuzzy. Results show that 83.9% of students are profiled as Good, 9.7% as Sufficient, 6.5% as Very Good, and 0% as Low. The model demonstrates that AQ contributes significantly to differentiated profiling: two students with identical assessment scores of 100 but differing AQ values (80 vs. 130) produced a 20.31-point gap in output values (Z* = 60.67 vs. Z* = 80.98). The model successfully captures continuous performance gradations that conventional single-score assessments cannot detect, providing more comprehensive, transparent, and pedagogically meaningful diagnostic information for learning intervention planning.