Claim Missing Document
Check
Articles

Found 2 Documents
Search

Instructor Performance Analysis in Educational Contexts Based on Learner Evaluation Data: Integration of Clustering and Predictive Model Lestari, Santi Dwi Desy; Margono, Hendro
MUKADIMAH: Jurnal Pendidikan, Sejarah, dan Ilmu-ilmu Sosial Vol 9, No 2 (2025)
Publisher : Prodi Pendidikan Sejarah Fakultas Keguruan dan Ilmu Pendidikan Universitas Islam Sumatera

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30743/mkd.v9i2.11772

Abstract

This study aims to analyze instructor performance in educational contexts by classifying instructors based on learner evaluation data through the K-Means clustering algorithm and developing a predictive model to support effective and targeted instructor development programs. The data were derived from learners’ evaluations of instructors, covering aspects such as discipline and professionalism, mastery of subject matter, and pedagogical skills in delivering content. The results indicate that k=3 is the optimal cluster, producing three categories: Superior Instructor, Potential Instructor, and Developing Instructor. Furthermore, the predictive model demonstrates that the Naive Bayes algorithm outperforms XGBoost in performance prediction, achieving higher accuracy, recall, precision, and F1-scores. The integration of clustering and prediction proves effective in enabling faster, objective, and data-driven decisions for instructor development. These findings provide significant implications for educational institutions in establishing adaptive and sustainable systems of instructor evaluation and management.‎
Mapping Sentiment towards Danantara: A Combined Clustering and Text- Based Predictive Model Lestari, Santi Dwi Desy; Yuadi, Imam
Journal of Law, Politic and Humanities Vol. 5 No. 6 (2025): (JLPH) Journal of Law, Politic and Humanities
Publisher : Dinasti Research

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38035/jlph.v5i6.2295

Abstract

Research aims to map public sentiment towards Danantara with the integration of clustering and text-based predictive models from social media data. Clustering using K-means obtained three clusters namely political criticism, neutral and prositive support. Linear SVM model performed best with 96% accuracy, followed by random forest (93%), Logistic Regression (90%) and Naïve Bayes (83%). The findings confirm that the public is highly sensitive to issues of transparency and governance in the establishment of Danantara, and the need for a responsive, data-driven public communication strategy. This research contributes to the public opinion monitoring system for national strategic policies.