University of Twente Student Theses
Using machine-learning models for operational exception handling : a case study at IBM
Schultz, Pim (2017) Using machine-learning models for operational exception handling : a case study at IBM.
|
PDF
2MB |
| Abstract: | Within service parts management, organizations need to maintain the balance between inventory levels and cost to prevent financial losses or customer dissatisfaction. The service parts management department of IBM has been managing inventory for almost a century and is continuously looking to improve their processes. Most of the standard ordering procedures within IBM have already been automated but human planners are still employed to handle exceptions. Given the progress made by IBM in cognitive computing systems, they are eager to apply their new technologies to existing process to improve their efficiency. Within this study, we research the possibilities of using cognitive computing systems in IBM’s exception handling process to improve its efficiency. We found that the characteristics of the operational exception handling process do not fit the strengths of a cognitive computing system and have therefore created a traditional machine-learning model. The machine-learning models are able to predict the planner’s actions in less than half of the cases. Given these results we advise against using the current model to automate the exception handling process but to use the output as a second opinion for the operational planners. We expect the performance of the model to be improvable by using a more diverse dataset, or by adding additional input variables. |
| Item Type: | Essay (Master) |
| Clients: | IBM, Amsterdam, Netherlands |
| Faculty: | BMS: Behavioural, Management and Social Sciences |
| Subject: | 50 technical science in general, 54 computer science |
| Programme: | Industrial Engineering and Management MSc (60029) |
| Link to this item: | https://purl.utwente.nl/essays/73979 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page