The Application of Machine Learning to Project Failure Prediction

Extracting Risk Triggers in Most Projects

Abstract

As system development projects become larger and more complex, quantitative project management becomes more important. In recent years, evolving management tools and development environments enables us to store enough project data. Using this data, we have tried predicting project failure applying machine learning, and confirmed that the accuracy has reached a level where it can be used in actual operations. On the other hand, it turns out that predicting failures is not enough to improve the project situation. This study aims at the early recovery of high-risk projects. We systematically extract risk triggers which lead many projects into trouble, and validate them by comparing with several cases studies.