Building on our successful development of a machine learning model for predicting reliable delivery time frames in parcel services, this thesis focuses on the critical next phase: expanding our dataset and validating model generalization across diverse user populations. Our existing model, developed in collaboration with the startup Green Convenience, has shown promising results on an initial student dataset. However, to ensure robustness and real-world applicability, we need to evaluate performance across broader demographic groups and larger datasets. This thesis will explore innovative approaches to user acquisition and conduct comprehensive generalization studies.