Article

A Comparative Study of Logistic Regression and Random Forest for Predicting Student Adaptability in Embedded Systems Entrepreneurship Education 

---

Abstract

The increasing demand for innovation in technology sectors necessitates a deeper understanding of the factors that foster adaptability among students in specialized fields like embedded systems entrepreneurship. This study provides a comparative performance analysis of two prominent machine learning algorithms—Logistic Regression and Random Forest—for predicting student adaptability levels (Low, Medium, and High). Utilizing a dataset comprising academic, experiential, and psychometric features, this research addresses the critical challenge of identifying student potential in a data-driven manner. The methodology employed a stratified data split and the Synthetic Minority Over-sampling Technique (SMOTE) to mitigate the severe class imbalance inherent in the dataset, followed by a rigorous hyperparameter tuning process for both models. The results revealed a nuanced outcome. While the linear Logistic Regression model achieved a superior overall accuracy (98.4%) compared to the more complex Random Forest model (87.2%), both algorithms completely failed to identify any instances of the 'High' adaptability class. This critical failure underscores the limitations of standard classification techniques when faced with extremely rare positive instances. Furthermore, a feature importance analysis conducted with the Random Forest model indicated that practical skills, such as innovation and model deployment scores, were the most significant predictors of adaptability, whereas traditional academic metrics like GPA had negligible influence. This study concludes that while AI-driven models show significant promise as an early-warning system to identify students who may require additional support, they are currently unsuitable for talent identification due to data limitations. The findings strongly advocate for a pedagogical shift in technical entrepreneurship education, emphasizing the need to prioritize experiential learning and practical skill development over conventional academic measures to cultivate the next generation of adaptable innovators.

Keywords: Adaptability, AI in Education, Embedded Systems, Predictive Modelling, Random Forest

How to Cite:

Siddique, Q., (2025) “A Comparative Study of Logistic Regression and Random Forest for Predicting Student Adaptability in Embedded Systems Entrepreneurship Education ”, Artificial Intelligence in Learning 1(3), 245-257. doi: https://doi.org/10.63913/ail.v1i3.34