Publicação
The role of explanations in artificial intelligence trust and comprehension: counterfactual vs. shap techniques
| Resumo: | This thesis extends the outcomes of the year-long Project-Based Learning initiative with NOS, a prominent telecommunications company in Portugal, focusing on optimizing the number of clients that should be flagged for specialized call center teams, to increase clients’ satisfaction. Two contributions improve model performance by addressing outlier management and applying ensemble techniques, each resulting in substantial improvements from initial solution. The remaining two focus on model explainability, including a deeper dive into the model’s outcomes and a study on how individuals interpret its explanations. Together, these studies complement the work done in PBL by improving both model performance and interpretability |
|---|---|
| Autores principais: | Penedo, João Pedro Lopes |
| Assunto: | Call center Time series forecasting Prediction modeling Explainability Explainability in practice AI adoption |
| Ano: | 2025 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | This thesis extends the outcomes of the year-long Project-Based Learning initiative with NOS, a prominent telecommunications company in Portugal, focusing on optimizing the number of clients that should be flagged for specialized call center teams, to increase clients’ satisfaction. Two contributions improve model performance by addressing outlier management and applying ensemble techniques, each resulting in substantial improvements from initial solution. The remaining two focus on model explainability, including a deeper dive into the model’s outcomes and a study on how individuals interpret its explanations. Together, these studies complement the work done in PBL by improving both model performance and interpretability |
|---|