A Designing a System to Predict Faculty Publication Performance Using a Time-Series Machine Learning Model (Case Study: Information Technology Study Program, PGRI University of Madiun)

Integrated Web-Based Forecasting System for Academic Publication Performance Evaluation Using ARIMA and SARIMA Models

Abstract

This research presents the development of an integrated web-based application designed to forecast and assess lecturers’ publication performance through time-series machine learning techniques. The study is conducted within the Informatics Engineering Program of Universitas PGRI Madiun by utilizing historical publication records as temporally ordered data. These records are transformed into structured time-series representations to reveal productivity trends, periodic patterns, and temporal variations in academic output. The system architecture consolidates data acquisition, preprocessing, forecasting, and visualization into a single platform implemented using the Django framework. Multiple time-series forecasting models are applied and evaluated to examine their predictive capability, with performance measured using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination (R²). The results indicate that the proposed system is capable of generating consistent and interpretable forecasts that reflect actual publication dynamics. By providing analytical insights and early trend detection, the system supports objective evaluation and assists academic management in strategic planning and continuous improvement of research productivity.