Uncertainty assessment in project scheduling with data mining

During project execution, especially in a multi-project environment project activities are subject to risks that may cause delays or interruptions in the baseline schedules. This paper considers the resource constrained multi-project scheduling problem with generalized activity precedence relations requiring multi-skilled resources in a stochastic and dynamic environment present in the R&D department of a home appliances company and introduces a two-phase model incorporating data mining and project scheduling techniques. This paper presents the details of Phase I, uncertainty assessment phase, where Phase II corresponds to proactive project scheduling module. In the proposed uncertainty assessment approach models are developed to classify the projects and their activities with respect to resource usage deviation levels. In doing so, the proposed approach enables the project managers not only to predict the deviation level of projects before they actually start, but also to take needed precautions by detecting the most risky projects. Moreover, Phase I generates one of the main inputs of Phase II to obtain robust baseline project schedules and identifies the risky activities that need close monitoring. Details of the proposed approach are illustrated using R&D project data of a leading home appliances company. The results support the efficiency of the proposed approach.