Skip to main content

Demand forecasting : Integrating Machine learning with experts judgment using Bayesian Networks

Primary supervisor

Mahdi Abolghasemi

Demand forecasting is the basis for a lot of managerial decisions in companies. During the last four decades, researchers and practitioners have developed numerous quantitative and qualitative demand forecasting models including statistical, machine learning, judgmental, and simulation methods. Several endogenous and exogenous variables can influence the dynamics of demand, and hence a single statistical model that only consists of historical sales data is often insufficient to produce accurate forecasts. In practice, the forecasts generated by baseline statistical models are often judgmentally adjusted by forecasters to incorporate factors and information that are not incorporated in the baseline models.

 Using expert judgment in complement to statistically analyzing large amounts of data has been shown to be beneficial for improving forecast accuracy. Moreover, evidence suggests that the human input to forecasts can be improved by providing a systematic approach to structure the information utilized when imposing judgment to make adjustments. However, when the number of peripheral contextual information increases, human factors can hinder judgment for reasons such as personal or social biases, heuristics, cognitive limitations, and system neglect.

 The question is more on how to appropriately process judgment and combine it with statistical/Machine learning forecasts to consistently improve the accuracy of forecasts. 

 Bayesian Networks (BN) are great tools to assess vast amounts of information, analyse their interaction, and combine qualitative with quantitative information for decision making. In this project, we aim to build a BN by combining the contextual information with historical quantitative information in a demand forecasting problem.

 The created BN can act like a Forecasting support system  (FSS) that ease the cognitive burden on the human mind and improves the accuracy of final forecasts by combining qualitative information (human judgment) and quantitative information.



Dalton, A., Brothers, A., & Walsh, S. (2013). Expert elicitation method selection process and method comparison. In Neuroscience and the Economics of Decision Making (pp. 202-214). Routledge.


Goodwin, P., & Wright, G. (2001). Enhancing strategy evaluation in scenario planning: a role for decision analysis. Journal of management studies, 38(1), 1-16.


Lawrence, M., Goodwin, P., O'Connor, M., & Önkal, D. (2006). Judgmental forecasting: A review of progress over the last 25 years. International Journal of forecasting, 22(3), 493-518.

Fildes, R., Goodwin, P., Lawrence, M., & Nikolopoulos, K. (2009). Effective forecasting and judgmental adjustments: an empirical evaluation and strategies for improvement in supply-chain planning. International journal of forecasting, 25(1), 3-23.

Required knowledge

Familiar with Bayesian Networks

Familiar with Forecasting