Article Per Year (5 Year)
p-Index From 2020 - 2025
0.751
P-Index
This Author published in this journals
All Journal REiD (Research and Evaluation in Education) BAREKENG: Jurnal Ilmu Matematika dan Terapan
Ibrahim, Zulfa Safina
Unknown Affiliation
Computer Science & IT Control & Systems Engineering Economics, Econometrics & Finance Education Energy Engineering Mathematics Mechanical Engineering Physics Transportation

Published : 4 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 4 Documents
Search

 1 
IMPACT OF FEATURE SELECTION ON DECISION TREE AND RANDOM FOREST FOR CLASSIFYING STUDENT STUDY SUCCESS Satiranandi Wibowo, Firdaus Amruzain; Retnawati, Heri; Sakti, Muhammad Lintang Damar; Khoirunnisa, Asma; Batubara, Angella Ananta; Berlian, Miftah Okta; Ibrahim, Zulfa Safina; Jailani, Jailani; Sumaryanto, Sumaryanto; Prasojo, Lantip Diat
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/barekengvol19iss3pp2083-2096

Abstract

The advancement of technology has a profound impact on the field of education. Education plays a crucial role in enhancing quality of life, particularly in higher education, where one of the key parameters is student success. This study investigates the influence of feature selection on the performance of machine learning models, particularly Decision Tree and Random Forest, in classifying student academic success. Utilizing a dataset of 19,061 students, the research aims to identify significant variables impacting classification outcomes. Feature selection was conducted using LASSO regression, resulting in a refined dataset of critical predictors. To address data imbalance, Synthetic Minority Over-sampling Technique (SMOTE) was applied, improving the representation of underrepresented classes. Both Decision Tree and Random Forest models were trained on balanced datasets, with performance evaluated using accuracy, precision, recall, and F1-score metrics. The Random Forest model demonstrated superior accuracy (96.41%) compared to the Decision Tree (67.15%), as well as higher AUC values. Model interpretability was enhanced using SHAP (SHapley Additive exPlanations). This study underscores the utility of advanced machine learning techniques in educational analytics, paving the way for data-driven decision-making to support student achievement.
Differential item functioning analysis of Arabic language exams across gender, study specialization, and geographic region in senior high schools Bakti, Anugrah Arya; Marzuki, Marzuki; Ibrahim, Zulfa Safina; Tuanaya, Rugaya; binti Yacob, Nur Yusra
REID (Research and Evaluation in Education) Vol. 11 No. 1 (2025)
Publisher : Graduate School of Universitas Negeri Yogyakarta & Himpunan Evaluasi Pendidikan Indonesia (HEPI)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21831/reid.v11i1.85961

Abstract

This study aims to examine the fairness of Arabic language assessment instruments used in Muhammadiyah senior high schools by detecting the presence of Differential Item Functioning (DIF) in the Final Semester Summative Test (UAS) for 12th-grade students in the Special Region of Yogyakarta during the 2023/2024 academic year. Using a descriptive quantitative design, the research analyzed student response data from 1,157 participants across 25 schools. Data collection was conducted through documentation of test blueprints, item sheets, answer keys, and student responses. Analysis was performed using the Lord and Generalized Lord methods within the framework of Item Response Theory (IRT), focusing on three demographic variables: gender, study specialization (science vs. social studies), and school region (Yogyakarta City, Sleman, Bantul, and Kulon Progo). The Rasch model was identified as the most optimal model due to its superior fit and fulfillment of key psychometric assumptions, including unidimensionality and parameter invariance. The findings indicate that several items exhibit significant DIF across all examined variables. Eleven items showed gender-based DIF, with a higher number favoring male students. Twenty-three items demonstrated DIF by study specialization, and thirty-seven items displayed DIF based on school region, with students from Yogyakarta City benefiting the most. These results suggest that the test is not fully equitable and highlight the need for item revision to ensure fairness. The study contributes theoretically to the field of educational measurement and practically to the development of fairer evaluation practices in Islamic and language education settings.
THE EFFECT OF SAMPLE SIZE ON THE STABILITY OF XGBOOST MODEL PERFORMANCE IN PREDICTING STUDENT STUDY PERIOD Damar Sakti, Muhammad Lintang; Jailani, Jailani; Retnawati, Heri; Hidayati, Kana; Waryanto, Nur Hadi; Ibrahim, Zulfa Safina; Khoirunnisa, Asma’; Satiranandi Wibowo, Firdaus Amruzain; Berlian, Miftah Okta; Batubara, Angella Ananta
BAREKENG: Jurnal Ilmu Matematika dan Terapan Vol 19 No 4 (2025): BAREKENG: Journal of Mathematics and Its Application
Publisher : PATTIMURA UNIVERSITY

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30598/barekengvol19iss4pp2679-2692

Abstract

Student success can be defined based on the period of study taken until graduation from college. Machine learning can be used to predict the factors that are thought to influence student success. To achieve optimal machine learning model performance, attention is needed on the sample size. This study aims to determine the effect of student sample size on the stability of model performance to predict student success. This research is quantitative. The data used is student data from a university in Yogyakarta from 2014 to 2019, totaling 19061 students. The target variable is the student study period in months, while the predictor variables are college entrance pathways, GPA from semester 1 to semester 6, and family socioeconomic conditions based on the father’s and mother’s income. This research uses the XGBoost model with the best hyperparameters and the bootstrap approach. Bootstrapping was performed on the original data by sampling twenty different sample sizes: 250, 500, 750, 1000, 1250, 1500, 1750, 2000, 2250, 2500, 2750, 3000, 3250, 3500, 3750, 4000, 4250, 4500, 4750, and 5000. The resulting bootstrap samples were replicated ten times. Model performance evaluation uses the Root Mean Square Error (RMSE) value. The result of this research is the XGBoost model with the best hyperparameters, obtained through the training data division scheme of 90% and testing data of 10%, which has the smallest RMSE value of 8.318. The model uses the best hyperparameters: n_estimators of 75, max_depth of 8, min_child_weight of 5, eta of 0.07, gamma of 0.2, subsample of 0.8, and colsample_bylevel of 1. The XGBoost model with optimal hyperparameters demonstrates peak performance stability at a sample size of 1750 students, as evidenced by consistent RMSE values across 10 bootstrap replications, confirming that this data quantity provides the ideal balance between prediction accuracy and stability for estimating study duration.
Stability of estimation item parameter in IRT dichotomy considering the number of participants Ibrahim, Zulfa Safina; Retnawati, Heri; Irambona, Alfred; Pérez, Beatriz Eugenia Orantes
REID (Research and Evaluation in Education) Vol. 10 No. 1 (2024)
Publisher : Graduate School of Universitas Negeri Yogyakarta & Himpunan Evaluasi Pendidikan Indonesia (HEPI)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21831/reid.v10i1.73055

Abstract

This research is related to item response theory (IRT) which is needed to measure the goodness of a test set, while item parameter estimation is needed to determine the technical properties of a test item. Stability of item parameter estimation is conducted to determine the minimum sample that can be used to obtain good item parameter estimation results. The purpose of this study is to describe the effect of the number of test takers on the stability of item parameter estimation with the Bayes method (expected a posteriori, EAP) on dichotomous data. This research is an exploratory descriptive research with a bootstrap approach using the EAP method. The EAP method is performed by modifying the likelihood and function to include prior information about the participant's 9 score. Bootstrapping on the original data is done to take bootstrap samples. with ten different sample sizes of 100, 150, 250, 300, 500, 700, 1,000, 1,500, 2,000, 2,500 were then replicated ten times and grain parameter estimation was performed. Each sample data with ten replications was calculated Root Mean Squared Difference (RMSD) value. The results showed that the 2PL model was chosen as the best model. The RMSD value obtained proves that many test participants affect the stability of item parameter estimation on dichotomous data with the 2PL model. The minimum sample to ensure the stability of item parameter estimates with the 2PL model is 1,000 test participants.
Co-Authors Alfred Irambona, Alfred Bakti, Anugrah Arya Batubara, Angella Ananta Beatriz Eugenia Orantes Perez, Beatriz Eugenia Orantes Berlian, Miftah Okta binti Yacob, Nur Yusra Damar Sakti, Muhammad Lintang Heri Retnawati Jailani Jailani Kana Hidayati Khoirunnisa, Asma Khoirunnisa, Asma’ Lantip Diat Prasojo Marzuki Marzuki Nur Hadi Waryanto Sakti, Muhammad Lintang Damar Satiranandi Wibowo, Firdaus Amruzain Sumaryanto Sumaryanto, Sumaryanto Tuanaya, Rugaya