Human-machine collaboration during replenishment planning – HUAWEI
|Company Name / Department||
Chung Hong Ng (Max)
Budapest, Hungary / The Netherlands
|Start Date||May 2022|
|Housing arranged by company||–|
Huawei is a leading global provider of information and communications technology (ICT) infrastructure and smart devices.
Sales offices across europe make decision on a demand plan according to pre-sales project and post-sales projects delivery plans, and to drive supply centers to stock up inadvance, however the demand plan are not accurate.
Planner in supply center has to adjust these plans from sales offices according to their own experiences and risk they identified, then place replenishment order to stock up in advance. This process is very labour intensive.
To make our process more efficient, we develop a system which consider multiple factors when generating replenishment plans and use different ML algorithms according to different items‘ characteristics to give out forecast plan.
However, the current system only uses ML models to give out forecast, other information that is not available in the system (e.g. the depreciation of the Turkish lira may lead to demand delayed) allows our planner to make better decisions compare to the system’s recommended output and this leads to output generated by the system through algorithms sometimes conflict with planner’s thoughts.
Problem to solve:
- Trust issues: Model output doesn’t align with planner’s experience causing planner not using model outputs.
2. Human-Machine collaboration: Currently most of our optimization focused on using constrains to find outcomes and automate base on that without taking the advantage of human decisions, and the result is not promising.
Goals of the Project
- Research on how to optimize the current model (maybe through algorithm improvement, model work flow improvement etc.) to give a more trusted output is necessary
- Taking advantage of human and machine made decision is crucial to make our model better. Therefore, research on how/where to integrate human decision with machine output is needed.
- To be discusses
Essential Student Knowledge
Operation Research background, knowledge in operation optimization, knowledge in linear and nonlinear optimization
More information: email@example.com