Article in Press
The articles listed on this page have undergone thorough peer review in accordance with the standards and policies of Enthusiastic: International Journal of Applied Statistics and Data Science and have been accepted for publication in forthcoming issues. However, they have not yet been assigned to a specific issue or given an official publication date. They will be formally published once included in a complete issue or volume on the journal’s website.
Volume 5 Issue 2, October 2025
Sentiment Analysis on the Rumors of the Discontinuation of IISMA Program using Support Vector Machine, Random Forest Classifier and XGBoost Classifier
Michelle Intan Handa a,1, , Maria Zefanya Sampe b,2*, Syafrudi c,3
a,b Business Mathematics, School of Applied STEM Universitas Prasetiya Mulya, Edu Town Kavling Edu I No. 1, Banten 15339, Indonesia
c Artificial Intelligence and Robotics, School of Applied STEM Universitas Prasetiya Mulya, Edu Town Kavling Edu I No. 1, Banten 15339, Indonesia
1 Email First Author; 2 [email protected]*; 3 [email protected]
Abstract
IISMA (Indonesian International Student Mobility Award) is a government-run student exchange program. Recently, rumors regarding its discontinuation have sparked various public opinions. This study aims to analyze these public sentiments and evaluate which machine learning model is most suitable for classifying sentiment labels in the dataset. The models tested include Support Vector Machine (SVM), Random Forest Classifier (RFC), and XGBoost Classifier. The dataset consists of 630 tweets scraped from Twitter and is split into an 80:20 ratio, with 80% allocated for training and 20% for testing. The results indicate that both SVM and RFC are the most effective models, achieving the highest accuracy of 85.44%. Sentiment analysis reveals that the majority of public opinion is positive, suggesting that most people agree with the discontinuation of the IISMA program. This is because the program is perceived as non-urgent and not a current national priority. These findings provide insights into public sentiment and highlight the utility of machine learning models in classifying such sentiment data effectively.
Spatio-Temporal Modeling of Crime Rates Using Geographically and Temporally Weighted Regression
Robiansyah Putra a,1,*, Fidelis Nofertinus Zai a,2
a Department of Mathematics, Faculty of Mathematics and Natural Sciences, Universitas Sumatera Utara, Jln. Bioteknologi 1, Medan, 20155, Indonesia
1 [email protected]*; 2 [email protected]
Abstract
This study analyzes the spatio-temporal modeling of crime rates in 35 regencies and cities in Central Java using the Geographically and Temporally Weighted Regression (GTWR) method. The objective is to investigate how socio-economic factors, including the open unemployment rate, percentage of the poor population, population density, average years of schooling, job vacancies, labor force participation rate, and labor wage, influence crime rates across different regions and periods. The goodness-of-fit test results indicate that the GTWR model has an R-squared value of 93.51%, higher than the 88.64% of the Geographically Weighted Regression (GWR) model, demonstrating GTWR’s ability to explain crime data variations that are heterogeneous both spatially and temporally. Partial significance tests and mapping results show that the influence of variables differs across years and regions, with population density and labor-related factors consistently being the main predictors. These findings highlight the importance of designing crime prevention policies that are locally tailored and based on spatio-temporal evidence.