Error modeling of demand patterns to improve forecasting accuracy
This study aims to estimate and model error patterns to reduce forecast error and improve forecast accuracy for time series data. The objective is to assess the impact of incorporating error patterns as features in long short-term memory and transformer neural network models. The research employs a...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/175223 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | This study aims to estimate and model error patterns to reduce forecast error and improve forecast accuracy for time series data. The objective is to assess the impact of incorporating error patterns as features in long short-term memory and transformer neural network models. The research employs a comprehensive approach, utilizing a dataset comprising over 450 time series to evaluate baseline and modified models. Results reveal mixed outcomes, with some instances showing enhanced performance while others demonstrate no significant improvement or decline in performance. These findings underscore the complexities inherent in error modeling and emphasize the need for further investigation. Despite inconclusive results, this study contributes valuable insights into the challenges and opportunities associated with error modeling in time series forecasting. This paves the way for future research endeavors aimed at refining and advancing error modeling techniques in this domain. |
|---|