Create a Knowledge and Data-Based Bayesian Network
In this project, we aim to explore the intricate connections between project management maturity and project operational performance, particularly concerning cost overruns. Both academic researchers and industry professionals often posit that higher project management maturity results in improved project outcomes. Empirical evidence in the literature supports this causal hypothesis, such as the significant reduction in cost overruns observed with increasing maturity levels.
To investigate this causal relationship, we’ll construct a sophisticated Bayesian network model to elucidate the causal and correlational links between project management maturity and the risk of project cost overruns. Bayesian networks employ observational inferences to derive conditional probability relations, providing a robust framework for our analysis. The proposed network will use project management best practices as input variables, while the output will be a probability range representing the likelihood of cost overruns for projects.
In this project, our objective is to develop a comprehensive causal model that integrates various sources of information, including expert knowledge of project management best practices and historical data on project performance. Project management auditors evaluate the implementation of best practices in client organizations and assess project outcomes. These experts possess valuable insights into causal patterns that can be used for diagnosis and prediction. However, clients often lack a well-structured database containing extensive project management maturity information for multiple projects within the same field, resulting in a limited number of well-documented cases.
Despite the challenges posed by incomplete data, we can leverage Bayesian networks to construct causal models with learning capabilities. Due to human computation limitations, asking experts to directly correlate numerous input variables (maturity evaluation) with various project performance criteria, such as cost overruns, is impractical. Instead, we can request that experts outline the most plausible structure for a causal network from a semantic perspective. The expert approach identifies dysfunctional aspects and implements preventive measures before addressing the damaging consequences.
Experts know about the underlying causes of failure (drift factors) and the causal relationship between project management maturity and the likelihood of encountering these factors. Therefore, our research approach or heuristic focuses on asking experts about the connection between project management maturity levels and drift factors rather than directly inquiring about the relationship between project management maturity and operational performance. This strategy enables us to develop a more robust and insightful causal model for the students of this course, providing them with a valuable learning experience.