Analyzing Student’s Perception of Attributes in Programming Assignment

Abstract

The increasing complexity of programming education necessitates a deeper understanding of student challenges and preferences to enhance learning outcomes. This study addresses this critical gap by exploring students’ perceptions of programming assignments using advanced computational methods. This research investigated three key questions: the major challenges students face in programming, their preferred methods for skill assessment, and the skills they associate with effective programmers. The methodology involved a survey of 508 participants, including Bachelors of Computer Application (BCA), Master in Computer Application (MCA), and alumni cohorts, utilizing mixed-format questions. Natural language processing techniques, including Term Frequency-Inverse Document Frequency (TF-IDF) with K-means clustering and VADER sentiment analysis, were employed to analyze the responses. Key findings revealed five distinct clusters of challenges with prevailing logic and syntax-related difficulties, a strong preference for error-explanation-based assessments over traditional methods, and prioritization of problem-solving and analytical thinking skills, accompanied by a neutral sentiment toward programmer attributes. These insights highlight the diverse obstacles that students encounter and the skills they value. These implications are significant for educational practice, suggesting tailored teaching strategies to address identified challenges and redesigning grading systems to incorporate interactive feedback mechanisms. This study contributes to the field by demonstrating the efficacy of natural language processing (NLP) and machine learning (ML) in educational data mining and offering a scalable approach to curriculum development. In conclusion, these findings lay the foundation for data-driven improvements in programming education with the potential to shape future pedagogical innovations and technology-enhanced learning environments.