Applying Neural Network Ensemble Concepts for Modelling Project Success

Researchers have been studying early planning process since the early 1990's and results from these researches suggest that projects with more early planning efforts are more likely to succeed. This study intends to employ neural networks to build credible models linking preproject planning and project success. Preproject planning status as measured by Project Definition Rating Index (PDRI) is set as the independent variable and schedule/cost performance is set as dependent variable. To enhance the performance of the neural networks model, bootstrap aggregation and boosting algorithms are incorporated in the model development process. The results from these two neural network ensemble models are examined. This research finds out that boosting neural network ensemble models produce better results than bootstrap aggregation neural network ensemble models. Results from both models show that project with better early planning can expect better chance of project success.