An SDG classification tool for South African research

Khutso Sepuru , Lamont Theron

Partner: sdg-hub

Year: 2020

Abstract: The existing South African SDG Hub's online tool seeks to be refined and augmented by the use of machine learning. Thus, various machine learning models were developed, trained and evaluated to classify South African research into multiple sustainable development goals whilst remaining cognizant of predictability, computability, and stability. External data was added to training data to improve data balance and classifier performance. Finaly, a production ready application developed using streamlit was deployed using logistics regression and random forest.

Presentation Video