Predicting Academic Achievement Categories in Senior High School: An Ordinal Logistic Regression Approach

Academic achievement is a key indicator of educational success and students’ readiness for higher education. This study employed Ordinal Logistic Regression (OLR) to identify the determinants of senior high school academic performance and to develop a predictive model for honor classification among students at Southern Baptist College, Cotabato, Philippines. The dataset comprised 884 students enrolled in three senior high school strands (ABM, HUMSS, and STEM). Descriptive results indicated a decline in academic performance during the pandemic period. Female students generally outperformed male students, while students from private schools tended to achieve higher honor distinctions than those from public schools. The final OLR model identified English, Science, and Filipino grades, type of previous school, and pandemic learning phase as significant predictors of honor classification. Among these, the pandemic learning phase showed the strongest effect (OR = 3.22, p < 0.001), indicating a substantially higher likelihood of attaining higher honors, whereas a public school background reduced the odds of achieving higher distinctions (OR = 0.39, p < 0.001). Model diagnostics indicated that the fitted model demonstrated an adequate goodness-of-fit and achieved approximately 65% predictive accuracy. Confusion matrix results revealed class imbalance, with the model predicting majority categories (“Without Honor” and “With High Honor”) more accurately than the extreme categories. Overall, the findings demonstrate the usefulness of Ordinal Logistic Regression in modeling ordinal educational outcomes and providing evidence to inform targeted educational interventions.