Klasifikasi Kesiapan Kerja Lulusan Universitas Dengan Pendekatan Data Mining Menggunakan KNN

Abstract

Work readiness is an essential indicator for assessing the quality of higher education graduates. This study aims to classify the work readiness of Informatics alumni at the Faculty of Computer Science, Universitas Singaperbangsa Karawang (UNSIKA), using the k-Nearest Neighbors (kNN) algorithm. The dataset was obtained from tracer studies of graduates from 2018 to 2024. The research follows the Knowledge Discovery in Databases (KDD) methodology, which includes five stages: data selection, preprocessing, transformation, data mining, and evaluation. To improve model performance, feature selection was performed using ANOVA, and data normalization was applied using the Standard Scaler. The classification model was tested under eight scenarios, with variations in the number of features, values of k, and training-testing data splits. The best result was achieved without feature selection, using a train-test ratio of 80:20 and k = 21, yielding 60.82% accuracy, 63.43% precision, 86.74% recall, and a 73.28% F1-score. The findings demonstrate that the kNN algorithm is suitable for predicting graduate work readiness, and that proper preprocessing and feature handling significantly affect classification performance.