Predicting the Success of Projects Using Evolutionary Hybrid Fuzzy Neural Network Method in Early Stages
Mehdi Fasanghari, Seyed Hossein Iranmanesh and Mohsed Sadegh Amalnick
The ability of predicting the project escalation at the early state of the project is a critical point for Information Technology (IT) projects. Undoubtedly, it is excellent for a project manager to control the project with both of time indexes and cost indexes simultaneously at the earliest stages. In this paper, we proposed a computational intelligence method based on Locally Linear Neuro fuzzy (LLNF) method for estimation and prediction of the time indexes and cost indexes of a project using Earned Value Management (EVM). Our proposed model is compromised with the traditional time series forecasting methods and artificial neural network method (multi-layer perceptron) for typical resource constrained project schedule problem (RCPSP). Results of the applied proposed method for predicting the success of Iranian IT projects show its accuracy, relevant, and applicability in IT project area.
Keywords: Earned value, project forecasting, Project time estimation, Locally linear neurofuzzy (LLNF) model, Locally linear model tree (LOLIMOT), Computational intelligence, ARIMA, Regression, Time series, IT project.