DataStage: Automated Machine Learning Platform

Abstract

Machine learning has become increasingly important across various domains, yet its complexity remains a significant barrier for non-expert users. Traditional ML workflows require extensive programming knowledge, understanding of statistical algorithms, and expertise in data pre-processing techniques, creating a substantial skills gap that prevents domain experts from leveraging ML capabilities. This paper presents DataStage, a comprehensive web-based automated machine learning platform designed to democratize access to machine learning by providing an intuitive, end-to-end solution for users without technical expertise. The platform features an interactive Vue.js frontend integrated with a Fast API backend powered by scikit-learn, offering numerous pre-processing tools like handling missing values, outlier handling, feature scaling, etc. DataSage supports both classification and regression tasks, providing intelligent algorithm recommendations based on dataset characteristics and problem types. The system includes visual analytics capabilities with real-time performance metrics, confusion matrices, and feature importance visualizations. Experimental evaluation on multiple benchmark datasets demonstrates that DataSage achieves comparable accuracy to manually optimized models while reducing development time significantly and eliminating the need for coding expertise.