Machine Learning & AI / Independent exploration
Delivery-risk Modeling
A delivery-risk modeling pipeline combining classical models, neural models, and explanation tools.
Problem
The repository explores prediction and explanation for delivery-related records.
Implementation
The main pipeline preprocesses data, trains multiple models including TabNet, evaluates predictions, and invokes SHAP and integrated-gradients explanations.
Results
VERIFIED: orchestration for these components exists in main.py. No corporate endorsement or independently validated predictive performance is asserted.
Limitations
The main pipeline passes X_test and y_test into the TabNet helper as its validation set, which is used for early stopping, then evaluates on those same records. This is not an untouched test estimate. Explanations do not establish causal factors.
Reproducibility
Reviewed at commit a85a7c663067 on 2026-11-09. Source inspection establishes implementation structure; repository-authored results are not independent measurements. The upstream application and benchmark suite were not rerun during this review.