Lowering Technical Barriers in Machine Learning Education: A Usability Evaluation of an Integrated Analytics Framework

Abstract

The increasing role of data analytics and machine learning (ML) across different fields has created a strong need for educational tools that support both theoretical understanding and hands-on practice. Although Python is widely used as the main programming language for data analytics, many students and
early-stage researchers find it difficult to use effectively because of steep learning curves in programming, statistics, and ML frameworks. This study introduces MLapi, an API-based machine learning tool designed to reduce technical barriers and make ML methods more accessible in educational and applied analytics settings. MLapi uses a three-tier architecture that links Microsoft Excel as the user interface with a Python-based analytics engine, allowing users to run statistical analysis and ML algorithms while viewing the underlying Python code in Jupyter Notebook format. To assess usability, an empirical study was conducted using the System Usability Scale (SUS) with data analytics professionals. The evaluation included reliability testing, hypothesis testing across demographic groups, and Principal Component Analysis to identify usability dimensions. Results show high overall usability, strong internal consistency, and no significant differences between demographic groups, indicating that MLapi provides an inclusive and user-friendly learning environment. These findings suggest that MLapi can serve as both an educational tool and a practical analytics solution, helping to create accessible ML learning environments and encouraging broader participation in data analytics education.
Keywords: Machine learning, Data Analytics, System Usability Scale, API-based Architecture, HumanComputer Interaction, Principal Component Analysis