Investigating Frank-Wolfe-based demand forecasting methods

With the growing emphasis on demand forecasting to ensure a company’s success and ability to make profits, it is imperative that businesses find a way to predict demand accu- rately to effectively allocate their resources and manpower. With the demand for demand forecasting predictive models, thi...

Full description

Saved in:
Bibliographic Details
Main Author: Soh, Chloe Sze Yee
Other Authors: Yan Zhenzhen
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2024
Subjects:
Online Access:https://hdl.handle.net/10356/175567
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:With the growing emphasis on demand forecasting to ensure a company’s success and ability to make profits, it is imperative that businesses find a way to predict demand accu- rately to effectively allocate their resources and manpower. With the demand for demand forecasting predictive models, this paper will explore further the novel Stochastic Subregion Frank-Wolfe (SSRFW) Algorithm. By adjusting different parameters to run simulations, this paper investigates how the SSRFW algorithm performs as a whole, as well as how its performance may be affected by its parameters. Further, this paper also looks to compare the SSRFW against a benchmark model to evaluate how well it does against other models. Overall, this study finds that the SSRFW algorithm consistently performs well under differ- ent scenarios. Though the effect of each parameter is examined during the investigation, it is noted that more in-depth analysis can be done in the future to complement the existing results.