MLapi: A Machine Learning API Tool for Data Analytics

Abstract

This study presents the design, development, and usability evaluation of MLapi, a novel machine learning API tool aimed at facilitating access to Python-based data analysis and machine learning techniques. MLapi was developed to address the growing need for user-friendly analytical tools that bridge the gap between technical complexity and accessibility, particularly for users with limited programming experience. The system architecture integrates Microsoft Excel as a front-end interface with a PHP-based API and a Python Anaconda back-end, forming a modular and scalable three-tier structure. MLapi offers pre-configured templates for statistical and machine learning methods, automatically generating results in Jupyter Notebook format to enhance transparency and educational value.
The empirical component of the study employed the System Usability Scale (SUS) to assess perceived usability among 150 data-analytics professionals in the Greek banking sector. MLapi achieved a mean SUS score of 90.0, exceeding the benchmark of 85 typically associated with excellent usability. Principal Component Analysis (PCA) of the SUS responses revealed three latent dimensions -Complexity, Agility, and Learnability- each demonstrating high internal consistency. Statistical analysis showed no significant differences in usability perceptions across gender, age, education level, or professional experience, indicating that MLapi provides a universally accessible user experience.
These findings suggest that MLapi is both technically robust and inclusive, offering intuitive workflows and minimizing cognitive load. The tool’s integration with Excel enhances accessibility, while its educational features support gradual skill development in Python and machine learning. The study contributes to the fields of educational technology and usability engineering by validating MLapi as a scalable and effective solution for data science education. Recommendations for future research include expanding the sample population, incorporating qualitative methods, and exploring integration with additional platforms to further enhance usability and applicability.