Estimating Risk in Information Technology Projects

Few studies have focused on estimating the relative effects of risk factors on IT project performance This paper discusses three predictive modeling techniques, regression, neural networks and classification trees, that could be used for such an estimate. A comparison suggests that classification tree techniques may provide more effective estimates than other techniques. A data set (n=227) with 9 risk factors is used to develop a classification tree estimate. Results from this analysis provide a classification tree with 8 nodes that demonstrates the ease of interpretation of the classification tree results. The results are relatively easy to understand and provide actionable information. While results are only preliminary, we conclude that a risk assessment tool based on classification tree techniques could provide a new and effective tool in the management of IT project risk.