CEU Electronic Theses and Dissertations, 2020
| Author | Szokolics, Daniel |
|---|---|
| Title | Revisiting estimated time of arrival models |
| Summary | Our client delivers pharmaceuticals to its clients on a daily basis, and we built an application for them which predicts the estimated time of arrival (ETA) of these goods. This application aims at enhancing customer satisfaction as the pharmacies can tell their clients when to come back for their drugs, and they can adjust their daily schedules accordingly. Before the application goes live, our client asked for an update in the prediction process in order to decrease the number of predictions with large errors. My capstone project was this prediction enhancement. Our client’s main request was to decrease the rate of large errors among the predictions (errors above 30 minutes). These errors added up to 15% of the total errors before the COVID-19 pandemic, and 25% during that. During the analysis I focused on the non-pandemic data, as I expected that the situation will get back to normal. There were two options to reduce these errors: 022; Fine- tune the underlying models • Drop some of the predictions which seem risky in advance |
| Supervisor | Szabolcs, Bíró |
| Department | Economics MSc |
| Full text | https://www.etd.ceu.edu/2020/szokolics_daniel.pdf |
Visit the CEU Library.
© 2007-2021, Central European University