Exploring the Impact of Collegiate Experiences on Academic Success and Retention for Non‐Traditional Students at Community Colleges: A Machine Learning Approach

This study examines the relationship between community college students’ demographic characteristics, collegiate experiences, and academic outcomes using data from the 2024 Community College Survey of Student Engagement (CCSSE). Guided by Astin's Input‐Environment‐Outcome (I‐E‐O) Model, the study explores how student engagement factors—such as active and collaborative learning, student effort, academic challenge, student‐faculty interaction, and institutional support—alongside demographic variables, relate to college GPA and total credit hours earned. The purpose is to compare traditional statistical and machine learning (ML) approaches in terms of their ability to predict academic success. Hierarchical regression and three ML algorithms—decision tree, Random Forest, and XGBoost—were applied to identify patterns and evaluate model performance. Engagement factors such as active learning and academic challenge were modestly associated with academic outcomes. While XGBoost performed best among the predictive models, the overall improvement in performance was limited. This study contributes to the literature by demonstrating how demographic and engagement variables can be jointly examined, and how ML, through the application of diverse algorithms and data‐splitting techniques, can serve as a complementary tool for testing research models and exploring variable influence, thereby enhancing confidence in interpreting results from large‐scale educational research. We focus on non‐traditional students in community colleges and clarify how CCSSE benchmarks map onto the constructs examined. We also discuss the practical interpretation of results and position supervised ML as a complementary tool that adds robustness via cross‐validation and out‐of‐sample testing.