Research map / Graph learning / Knowledge and temporal graphs
Deep Hierarchical Graph Convolution for Election Prediction from Geospatial Census Data
Mike Li, Elija Perrier, Chang Xu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use graph convolution over SA1 neighbours instead of Random Forest | Election 2PP prediction · 80% train | RMSE: 7.75 -> 2.37 | Improved |
| add hierarchical SA2 graph (V2) to standard GCNN | Graph-CNN · 80% train | RMSE: 2.37 -> 2.20 | Improved |
| add hierarchical SA2 graph (V2) to standard GCNN | Graph-CNN · 20% train | RMSE: 3.20 -> 2.81 | Improved |
| add hierarchical SA2 graph (V5) to standard GCNN | Graph-CNN · 10% train | RMSE: 3.94 -> 3.39 | Improved |
| use more complex variants (V4/V5) instead of simple projection V2 (ablation) | Hierarchical GCNN · 80% train | RMSE: 2.20 -> 2.69 | Hurt |
| use more complex variant V5 instead of V2 with 10% training data (ablation) | Hierarchical GCNN · 10% train | RMSE: 3.45 -> 3.39 | Improved |
| add Mahalanobis distance learning to projection edges (ablation) | Hierarchical GCNN · 80% train | RMSE: 2.20 -> 2.15 | No clear effect |
| add Mahalanobis distance learning to projection edges (ablation) | Hierarchical GCNN · 10% train | RMSE: 3.45 -> 3.43 | No clear effect |
| add third SA3 level to V5 (ablation) | Hierarchical GCNN (V5) · 10% train | RMSE: 3.39 -> 3.28 | Improved |
| add third SA3 level to V2 (ablation) | Hierarchical GCNN (V2) · 80% train | RMSE: 2.20 -> 2.52 | Hurt |
| add third SA3 level with graph convolution to V2 (ablation) | Hierarchical GCNN (V2) · 10% train | RMSE: 3.45 -> 3.37 | Improved |
| graph convolution at SA2 level instead of direct projection (ablation) | Hierarchical GCNN · all splits | RMSE | Improved |
Closest papers by meaning
- Graph-Time Convolutional Neural Networks: Architecture and Theoretical Analysis (TPAMI 2023 · similarity 0.85)
- Spatial-Temporal Synchronous Graph Convolutional Networks: A New Framework for Spatial-Temporal Network Data Forecasting (AAAI 2020 · similarity 0.84)
- Multi-GCN: Graph Convolutional Networks for Multi-View Networks, with Applications to Global Poverty (AAAI 2019 · similarity 0.84)
- Dynamic Spatial-Temporal Graph Convolutional Neural Networks for Traffic Forecasting (AAAI 2019 · similarity 0.83)
- Hierarchical Graph Convolution Network for Traffic Forecasting (AAAI 2021 · similarity 0.83)
- pathGCN: Learning General Graph Spatial Operators from Paths (ICML 2022 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
