Claim Missing Document
Check
Articles

Found 5 Documents
Search

SURVIVAL ANALYSIS ON DATA OF STUDENTS NOT GRADUATING ON TIME USING WEIBULL REGRESSION, COX PROPORTIONAL HAZARDS REGRESSION, AND RANDOM SURVIVAL FOREST METHODS Rachmawati, Ramya; Afandi, Nur; Alwansyah, Muhammad Arib
BAREKENG: Jurnal Ilmu Matematika dan Terapan Vol 19 No 3 (2025): BAREKENG: Journal of Mathematics and Its Application
Publisher : PATTIMURA UNIVERSITY

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30598/barekengvol19iss3pp2111-2126

Abstract

This article presents a comprehensive study of the factors that influence the length of study data of undergraduate students at FMIPA UNIB class 2018 and 2019. This study is essential because observations show that many students study for more than 8 semesters. The purpose of this study is to determine the factors that significantly influence the length of study of undergraduate students. These factors can be internal and external. Survival analysis is the right method to identify these factors because ordinary regression analysis is unable to estimate survival data. Therefore, methods such as Weibull regression, Cox Proportional Hazards regression, and Random Survival Forest are used. This study does not compare the methods used because these methods are independent of each other, but have the same goal, namely, to determine the factors that influence the length of study of students. The data used in this study are data on the length of study of students from the 2018 and 2019 cohorts sourced from the academic subsection of FMIPA UNIB, with variables of GPA, gender, region of origin, university entry route, parents' occupation, type of study program, and length of study. The results showed that GPA and the type of study program significantly influenced the length of study in Weibull regression analysis. In Cox proportional hazard regression, the GPA variable is an influential factor, while using the Random Survival Forest method, all factors significantly influenced the length of study, with their respective levels of importance.
The Disparity of Maternal and Neonatal Death Modeling in Sumatra Region Using Geographically Weighted Bivariate Negative Binomial Regression Bayubuana, Muhammad Gabdika Bayubuana; Nugroho, Sigit; Rini, Dyah Setyo; Alwansyah, Muhammad Arib
Journal of Statistics and Data Science Vol. 3 No. 2 (2024)
Publisher : UNIB Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33369/jsds.v3i2.41285

Abstract

The Sumatra region occupies the second highest rank in terms of Maternal Mortality Rate (MMR) and Neonatal Mortality Rate (NMR) in Indonesia in 2020. Many factors are thought to have influenced these two cases, both directly and indirectly. So it is necessary to do an analysis to find out what factors influence MMR and NMR. The methods that can be used to determine these factors are Bivariate Negative Binomial Regression (BNBR) and Geographically Weighted Bivariate Negative Binomial Regression (GWBNBR). The results of the analysis show that the Deviance Information Criterion (DIC) in GWBNBR is smaller than BNBR, so GWBNBR is better than BNBR in modeling MMR and NMR in the Sumatra Region in 2020.
A Panel Data Spatial Regression Approach for Modeling Poverty Data In Southern Sumatra Hidayati, Nurul; Karuna, Elisabeth Evelin; Alwansyah, Muhammad Arib
Journal of Statistics and Data Science Vol. 3 No. 2 (2024)
Publisher : UNIB Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33369/jsds.v3i2.41288

Abstract

This research examines the use of spatial panel data regression approach to model poverty data in the Southern Sumatra region. The main objective of the study is to model poverty in the Southern Sumatra region using spatial panel data regression. Panel data from districts/cities in South Sumatra, Jambi, Lampung, Bengkulu, and Bangka Belitung during the 2015-2021 period were used in the analysis. The spatial panel models used in this study are panel SAR regression and panel SEM. The results show that the spatial panel data approach is better at explaining variations in poverty levels compared to non-spatial models. A significant spatial spillover effect was found, where the poverty level of an area is influenced by the conditions of its neighboring areas. The results of the analysis show that the best model to use in modeling the Poverty Percentage data in the Southern Sumatra region is the Spatial Autoregressive Fixed Effect (SAR-FE) model based on the smallest AIC and BIC values. Factors such as average years of schooling and life expectancy are proven to have a significant influence on the percentage of poverty in the SAR Fixed Effect model.
MODELING THE MANY EARTHQUAKES IN SUMATRA USING POISSON HIDDEN MARKOV MODELS AND EXPECTATION MAXIMIZATION ALGORITHM Alwansyah, Muhammad Arib; Rachmawati, Ramya
BAREKENG: Jurnal Ilmu Matematika dan Terapan Vol 18 No 1 (2024): BAREKENG: Journal of Mathematics and Its Application
Publisher : PATTIMURA UNIVERSITY

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30598/barekengvol18iss1pp0163-10135

Abstract

Sumatra Island is one of the islands that are prone to earthquakes because Sumatra Island is located at the confluence of three plates, namely the large Indo-Australian plate, the Eurasian plate and the Philippine plate. In general, the number of earthquake events follows the Poisson distribution, but there are cases where there is overdispersion in the Poisson distribution. The Poisson Hidden Markov Models (PHMMs) method is used to overcome overdispersion, then applying the Expectation-Maximization Algorithm (EM algorithm) to each model to obtain the estimated parameters. From the models obtained, the best model will be selected based on the smallest Akaike Information Criterion (AIC) value. The data used is secondary data on earthquake events on the island of Sumatra from January 2000 to December 2022 with a depth of ≤ 70 Km and a magnitude of ≥ 4.4 Mw. From the research, the model with m = 3 is the best estimation model with an AIC value of 1503,286. From the best model, estimates are obtained for Poisson Hidden Markov Models with an average occurrence of earthquakes of 5.7633 ≈ 6 events within one month.
Sentiment Analysis of Twitter User’s Perceptions of the Campus Merdeka Using Naïve Bayes Classifier and Support Vector Machine Methods Salsabilla, Intan; Alwansyah, Muhammad Arib; Nugroho, Sigit; Agwil, Winalia
Journal of Statistics and Data Science Vol. 2 No. 2 (2023)
Publisher : UNIB Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33369/jsds.v2i2.30577

Abstract

The Campus Merdeka program is being implemented by the government to realize autonomous and flexible learning in tertiary institutions to create a learning culture that is innovative, not restrictive, and the needs of students. The Campus Merdeka provides added value and is attractive and provides various responses from the public both directly and on different social media platforms. One of the social media platforms is Twitter. Therefore, research was conducted on the community's response to the Campus Merdeka program on Twitter social media. Twitter documents in the form of community response tweets to the Campus Merdeka program are classified into two categories, namely positive responses and negative responses. The method used in this study is the Naïve Bayes Classifier (NBC) and Support Vector Machine (SVM) with a Polynomial Degree 2 kernel. The highest level of accuracy resulting from this research is 73.5% with a parameter value of  of 0.5, a constant value  is 0.5, with training data of 309 documents for training data and 132 documents for test data. The accuracy results obtained for the Naïve Bayes Classifier method are 65.9% and for the Support Vector Machine method, an accuracy is 73.5%.