Master Thesis: Demand Forecasting

Join our Team

About this opportunity:

Accurate demand forecasts are essential for effective supply-chain planning. They support decisions concerning inventory, capacity, procurement, production, transportation, and customer service across different time horizons. However, forecast accuracy should not be assessed in isolation. A forecast that is statistically more accurate does not necessarily generate better business outcomes if it requires disproportionate effort, creates unnecessary complexity, or does not improve decisions elsewhere in the supply chain.

Different products and customer segments may require different forecast horizons, levels of granularity, forecast methods, and sources of expert opinion. Statistical, machine-learning, driver-based, manual, tender, and one-time forecasts may each be appropriate in different circumstances. Forecasting therefore needs to be managed as an integrated framework covering forecast objects, horizons, methods, opinions, segmentation, business logic, business rules, and performance metrics. The framework should enable fact-based decisions about where forecasting effort creates value and where additional effort is unlikely to provide a positive return.

What you will do:

The thesis will investigate how demand forecasting performance can be evaluated and managed based on its total business impact, rather than forecast accuracy alone. The study should develop a conceptual and practically applicable framework for determining:

• Criteria for how forecast methods, opinions, granularity, and review effort should be differentiated across product-customer segments and time horizons

• Which performance metrics are suitable to govern the criteria for different product-customer segments and time horizons

• How to identify the point at which improving forecast accuracy further costs more than the resulting business benefit

The research may combine a literature review with empirical analysis of historical forecasting and supply-chain performance data. Relevant methods may include statistical analysis, forecast-error decomposition, segmentation analysis, cost-benefit analysis, simulation, and interviews with supply-chain and business stakeholders. The final framework should support decisions regarding where to increase, maintain, automate, or reduce forecasting effort.

The skills you bring:

• Suitable background: Supply Chain Management, Industrial Engineering, Operations Research, Data Science, Business Analytics, Economics, or a related field


• Knowledge of forecasting, statistics, optimization, and data analysis is advantageous