Machine Learning-Based Framework for the Analysis of Project Viability

Abstract

We constructed a machine learning based analytical framework for transforming the project data of African Development Bank (AfDB) into actionable insights and for uncovering hidden business opportunities offered by markets in Africa. This framework helps identify skyrocketing sectors, emerging markets and fast-growing economies that have a massive impact in shaping the future of the continent. As a result of the foregoing, this has been implemented as an approach which predicts where investment needs are necessary. Furthermore, we collected a dataset containing more than 1,400 projects, where 92.9 percent of them have project descriptions and are relatively well-documented. We then loaded them into a single corpus for extracting tiny details relating to the business opportunities and giving investment directions to follow based on the promising sectors. We used Random Forests and Latent Dirichlet Allocation respectively to classify the most fruitful sectors for investments and to derive meaningful topics that potential investors might consider when investing in the continent.