THE IMPACT OF AI-DRIVEN EDUCATIONAL ANALYTICS ON PREDICTING ACADEMIC CHANGES AND REDUCING STUDENT DROPOUT RATES: A FIVE-COUNTRY SIMULATION FRAMEWORK FOR JORDAN, THE UAE, THE USA, TÜRKIYE, AND SAUDI ARABIA

Artificial intelligence (AI)-driven educational analytics can provide institutions with early-warning information about academic progression and possible student dropout. This study evaluates machine-learning prediction of student outcomes and develops a transparent five-country contextual simulation covering Jordan, the United Arab Emirates (UAE), the United States (USA), Türkiye, and Saudi Arabia. The empirical component uses 4,424 real student records from the UCI Predict Students' Dropout and Academic Success dataset, containing 36 predictive features and three outcomes: Dropout, Enrolled, and Graduate. Logistic Regression and Random Forest models were evaluated using a stratified 80/20 holdout sample. In addition to predictive metrics, inferential analyses were conducted using one-way ANOVA for continuous variables and chi-square tests for categorical variables, with effect sizes and adjusted pairwise comparisons reported. The results show significant differences among outcome groups across multiple academic-performance variables, while several categorical characteristics are also significantly associated with outcome. Random Forest achieved higher holdout predictive performance than Logistic Regression in the present experiment. A World Bank contextual layer was used for the five participating countries; because the UCI student records were not sampled from those countries, country assignment is explicitly synthetic and is not interpreted as observed student nationality, residence, or institutional location. The study therefore provides a reproducible simulation framework rather than national dropout estimates. The findings support the use of AI as a decision-support component of early-warning systems, while emphasizing that prediction accuracy alone cannot establish dropout reduction. Longitudinal intervention evidence, external validation, fairness assessment, and country-specific student data are required before causal or national claims can be made.