Hierarchical Evaluation of Advanced Time Series Forecasting Models for Dynamic Demand Prediction
Abstract
Demand forecasting is an indispensable business tool for an E-commerce business to effectively plan and make resource utilization decisions. A comparative analysis on five advanced demand forecasting models: DeepAR, N-BEATS, N-HiTS, TimesNet, and LSTNet in forecasting the printing shops demand by using prior years data. Each was tested on their ability of pattern and trend recognition of temporal evolution. From the comparison of these five models, DeepAR outperforms with the highest accuracy. This study will guide to the selection of the right model for time series forecasting tasks and help in better decision making in resource management.
Fields of Study & Subjects
Forecasting Techniques and Applications
Time Series Analysis and Forecasting
Energy Load and Power Forecasting
Publication details
- Journal / source
- CRC Press eBooks
- Pages
- 293–302
- Publication date
- 25 February 2026
- Language
- EN
- Document type
- Book chapter