Autonomous Data Scientist Agent: A Multi-Agent System for End-to-End Machine Learning Automation

Considering the fact that at present we live in an era of digitization, when data plays a crucial role in various aspects of life, from education to business, from medicine to research, it looks vital to create some mechanisms that would assist in processing the information provided quickly and efficiently. On the contrary, conventional approaches to data science always require professionals who will carry out the analysis and interpret the obtained results. This paper aims to explain the Autonomous Data Scientist Agent – a web-based mechanism which will be using advanced technologies of machine learning, conversational AI, and data visualization to enable users to process their datasets, perform automatic data analysis, examine machine learning models and obtain valuable information. The difference between the existing solutions and this web application lies in the fact that here people will have an opportunity to communicate in a natural way and get notified instantly. However, problems related to automation, accessibility, and interpretability become the major ones when considering the system under discussion, which guarantees automatic processing, analyzing, modeling, and predicting due to decreased participation of people in these processes. The current research paper contains an overview of the developed web-application architecture along with information about the method of its development.