نبذة مختصرة : This dissertation aims to understand the academic outcome disparity between underrepresented minorities in higher education when compared to other racial groups. It seeks to address the social inequities in learning, college integration, and completion rate. The focus was narrowed to a specific marginalized community that represents first-generation underrepresented minority (FGURM) students, that is, students whose parents have not obtained a post-secondary degree and identified as belonging to the following racial or ethnic group: Blacks, Hispanics/Latinos, American Indians/Alaska Natives, Native Hawaiian/Pacific Islanders, and two or more races in the United States. The overall objective was to explore with predictive models how demographic factors, pre-college academic performance, socioeconomic status, targeted programs aimed at fostering integration into campus communities, and support systems can increase the likelihood of academic success within this group. Predictive models based on supervised machine learning algorithms like Random Forest in combination with ensemble learning techniques like bagging and boosting was used to assess various predictors of successful academic outcomes. To address issues like incomplete and imbalanced data, a combination of case deletion and imputation methods, such as K-Nearest Neighbor (KNN), Linear Regression, and the Synthetic Minority Over-sampling Technique - Edited Nearest Neighbors (SMOTEENN), were utilized. The results suggested that pre-college academic achievements, assessed through standardized test scores (ACT/SAT), along with demographic factors such as age, gender, and ethnicity, are significant predictors of cumulative grade point average (CGPA). Furthermore, a combination of test score and CGPA was identified as a strong predictor of graduation outcome. The research further showed that student involvement particularly in academic related organizations is vital for academic achievement. Other forms of student involvement, such as participation in cultural ...
No Comments.