DATA SCIENCE – Urban Delivery Demand GAT Model – 100% Complete

[snimle123]

Week 5 pull request for final submission

Use case includes:

• Urban delivery demand prediction using Graph Attention Networks (GAT)
• Data preprocessing and feature engineering
• Time-based feature extraction and zone creation
• Traffic level and driver availability encoding
• Graph construction using zone relationships
• GAT model implementation using PyTorch Geometric
• Model training, evaluation, and visualisation
• Actual vs predicted demand analysis
• MAE, RMSE, and R² performance evaluation
• Future improvements and model optimisation discussion
• Dataset included for experimentation and reproducibility

[nadunhs25]

Great work, keep it up, guys

[PriyanshuCauleechurn]

Reviewed the PR and overall the notebook is well organised and easy to follow. The workflow from preprocessing through to modelling and analysis is structured clearly, and the markdown explanations help improve readability throughout the notebook. The implementation and overall presentation of the use case are also well done.

I also checked the project structure and exported files, and everything appears to be organised properly from my side. The notebook sections are separated clearly, making the analysis process easier to understand and review.

One thing that could still be improved is adding a proper Scenario section along with Duration, Level, and Pre-requisite Skills to make the notebook more complete and aligned with the final publishing format used across the project.

Apart from that, the overall work looks good from my side.

Approved with minor suggestions.

[Litxinh123]

Hi snimle123, looks good overall!

• The preprocessing steps are clear, including missing value handling, datetime conversion, and feature creation.
• The graph construction flow is suitable for the GAT approach, and the model training/loss visualisation also looks good.
• The actual vs predicted comparison helps show the model behaviour clearly.

However, a few things could still be improved:

• The use case name/file name should be updated to follow the required format, e.g. UCxxxxx_Urban_Delivery_Demand_GAT_Model.
• The notebook seems to be missing the template section (Authored by, Duration, Level, Pre-requisite Skills).
• It would be helpful to add a bit more explanation after the evaluation results and prediction outputs.
• Kindly include all required submission files, such as .json, and .html, for a more complete handover package.

Kindly let me know if there's any concern. Thanks

[snimle123]

Hi reviewers,

Thank you for the feedback. I addressed the requested changes and pushed an updated version of the notebook approximately two weeks ago.

The updates include:
• Additional documentation and explanations
• Results interpretation section
• Future improvements section
• Project handover information
• Other requested notebook improvements

Could you please review the latest version when convenient and let me know if any further changes are required?

Thank you for your time and feedback.

[PriyanshuCauleechurn]

Reviewed the updated notebook and the requested changes have been addressed well. The additional documentation, results interpretation, future improvements, and project handover sections provide much more context and improve the overall readability of the notebook.

The workflow is now easier to follow, and the added explanations help readers better understand both the methodology and the outcomes of the project. The notebook is well organised and aligns more closely with the expected publishing standard.

Overall, the notebook is in good shape and I do not have any further major concerns. Nice work addressing the feedback and improving the quality of the submission.