Non Linear EVM based on Support Vector Regression Growth Model for Predicting Project Completion Time

Abstract

Earned Value Management (EVM) is a method used to monitor and to predict project completion time. This method uses a linear approach in predicting project time completion. Unfortunately, most of the projects run in dynamic environments with complex characteristics, causing project progress to require a non-linear approach. That is why the use of EVM in monitoring and predicting non-linear project completion time is less effective. This study proposes a more realistic alternative approach using non-Linear EVM based on the Support Vector Regression (SVR) - Growth Model. The SVR-growth model is used to accommodate the non-linear progress of the project, while the EVM is used to represent the predicted results of the project completion time. For model validation, 5 data on oil and gas field development construction projects in Jawa, Bali and Nusa Tenggara Regions were used as case studies. The results of this study indicate that the results of project completion time prediction using the SVR-Growth Model provide high accuracy and precision compared to the traditional EVM method.

Keywords

• support vector regression
• non-linear growth model
• earned value management.