| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 742.15 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
This thesis examines applying Graph Neural Networks (GNNs) to enhance U.S.
dealership network planning for a luxury car manufacturer. A literature review on location
factors for car dealerships informs the collection of 65 county-level explanatory variables
across five categories. An ablation study on 34 variable combinations and ten state-of-the-art
GNN operators ranks the variable categories by their predictive capabilities. The results reveal
the significance of competition presence, demographics, and wealth in location decisions and
propose seven counties as potential expansion locations. Overall, this research demonstrates
the applicability of GNNs in addressing complex spatial decision problems, offering insights
for industry practitioners and researchers.
Descrição
Palavras-chave
Business analytics Graph neural networks Applied data science Location intelligence
