This is the group project from my Graduate Project Development Class STSCI5954

We tried to create a model to analyzes data from 777 U.S. colleges and universities to explore the impact of institutional characteristics, such as public/private status and student services, on key performance metrics like application rates, graduation rates, and alumni donations. Employing exploratory data analysis and predictive modeling, the study classifies institutions as elite or non-elite. Random Forest, the best-performing model, achieves superior accuracy and ROC AUC compared to Logistic Regression. Key insights include the significance of out-of-state tuition, academic reputation, and student support services in enhancing institutional outcomes.

The key question of this project is: How do institutional characteristics such as public/private status, academic reputation, and student services impact key performance metrics, including application rates, graduation rates, and alumni donations, for U.S. colleges and universities? Based on the key performance metrics, we are able to classify institutions into elite or non-elite categories to evaluate their outcomes more effectively. This analysis is crucial because it provides actionable insights that have the potential to directly benefit students, enhance institutional strategies, and positively impact the broader education sector.

To address this question, we analyze a dataset of 777 U.S. colleges and universities. This analysis involves exploratory data analysis (EDA), predictive modeling, and the evaluation of relationships between variables. We leverage EDA to uncover patterns and relationships within the dataset. In addition, we use predictive modeling techniques to distinguish elite institutions and quantify the factors contributing to their success. Random Forest is identified as the best-performing model, outperforming Logistic Regression in both accuracy and ROC AUC, indicating its superior ability to capture complex relationships between features and target variables.