Investigating Aptitude in Learning Programming Language Using Machine Learning and Natural Language Processing

Abstract

This study investigates the relationship between prerequisite courses and skill acquisition in programming education. It proposes a case study examining cognitive, natural language, and mathematical aptitude indicators as predictors of programming performance. Analyzing data from 1238 undergraduate students at Riphah International University, the research employs Machine Learning models to predict outcomes, achieving high R2 scores and low Mean Squared Error rates. A zero-shot text classification model identifies required aptitude skills: 62% cognitive, 24% natural language, and 14% mathematical. These skills are mapped to predicted programming course scores, offering a new approach to understanding programming language aptitude. The study aims to bridge the gap between prerequisite courses and subsequent skill development, contributing valuable insights to computing education curriculum design.