Meeting notes week 5 - davidlabee/Graph4Air GitHub Wiki

🗓️ Meeting Notes – 29 April 2025, 14:00

🗒️ Last Meeting’s Feedback

Feedback for Pieter

  • Confirm current aggregation method “works well enough” for initial experiments.
  • Check the stats (distribution) of the current aggregation and try other options.
  • Think about alternative graph construction with Highway segments separate from the rest.

Feedback for David

  • Parallelize the similarity-computation function.
  • Instead of leaving nodes out, try to compute similarity across all node pairs.

General Notes

  • Continue refining baseline GAT/GCN architectures and document choices in the wiki.
  • External validation: Use Palmes tube data for final validation—and possibly include in training.
  • Evaluation Metrics: Report PearsonR, MAE and RMSE.
  • Dataset Imbalance: Highway segments dominate; consider penalizing over-represented classes in the loss.

🗒️ Progress since last meeting

Pieter

  • Added all features to the baseline models.
  • Created function for handling missing target(NO2) values in baseline graph construction -> Baseline Graph Design
  • Organized all the code in different notebooks
    • Data exploration
    • Baseline Graph
    • Basic GNN models
    • Multi Resolution model
  • updated wiki -> Baseline Model Architecture
  • first trials with multi-scale GNN look promising
  • created scoreboard -> Scoreboard

David

  • Created a little different graph augmentation method
  • Will present the first testing results and working of the new graph augmentation method

🗒️ New Meeting Notes

General Notes

  • Check the unit for NO2d data. Is it the same as in the research of the supervisors?
  • Should we use a mask for the missing values instead of mean imputation + dropping?
  • What is the effect of the current imputation function on performance?
  • Put the created Colab notebooks on Github as well.
  • Fill in the Scoreboard page to document results.
  • Try the cross validation to see if there is overfitting.
  • Think about bias, overfitting and look at ways to prevent these.
  • Also consider other GNN python packages.
  • Do external validation on the Palmes dataset for all models. (could maybe be done during hyperparameter tuning)
  • If you are satisfied with the current graph structure look into hyperparameter tuning
  • Is there is a way to take 'half' of the Palmes measurements into the graph structure (for training)

Feedback for Pieter

  • Refine basic aggregation method using existing graph partitioning algorithms.
  • Optional: try multiscale graphs where aggregation happens based on road types.

Feedback for David

  • For next week create some good comparisons of the graph augmentation model with different parameters, scores and the baseline model scores.

Next Meeting: TBD