Identification of Risk Factors in Globally Outsourced Software Projects using Logistic Regression and ANN

Abstract

Abstract— Global supply chains are often critically dependent upon globally outsourced Information Technology (IT) and Business Process Outsourcing (BPO) projects.  Effective risk assessment and mitigation of these projects is therefore of great importance for such supply chains. Traditional risk management techniques used so far in IT and BPO projects have depended almost entirely on the ‘Expected Utility Theory’ that computes risk exposure as the product of risk probability and risk impact. Although this method is considered the gold standard in risk assessment, it has severe limitations due to the fact that accurate computation of risk probability and impact is difficult.  In this and earlier papers, we have advocated the use of ‘risk factors’ in conjunction with other existing methods for risk assessment. Risk factors are conditions that affect project performance. In this paper, we use Logistic Regression and Artificial Neural Network (ANN) based methods on a set of globally outsourced projects to predict project performance and to rank the relative importance of project risk factors. Although in this paper these techniques have been tested in outsourced IT projects, these can also be used in identifying and determining project risk factors in other industries.

Keywords— Outsourced projects, Risk Management, Risk Factors, Logistic Regression, Artificial Neural Networks