Case 3 Ablation Study Results for Graph-Based Power Load Forecasting

Mengze Gu published results from an ablation study on figshare in April 2026. The dataset contains performance comparisons for an Edge Graph Attention Network (EGAT) model against traditional methods like GRU, MLP, and LSTM for power load forecasting. The file is a 5.5 KB Excel spreadsheet.

Use Cases

• Benchmarking graph-based forecasting models against traditional time-series methods based on the described model comparisons.
• Analyzing the impact of graph construction on forecasting accuracy based on the described connectivity patterns and edge features.
• Evaluating the trade-off between model performance and computational cost in load forecasting based on the described limitations.
• Studying feature interaction representation in time-series data based on the described node and edge feature combination method.

Strengths

• Dataset is openly licensed under CC-BY-4.0.
• File size is 5.5 KB, indicating a concise and focused set of results.
• Results are from a recent study, last updated in April 2026.

Limitations

• Row count and specific column definitions are unknown, limiting suitability assessment.
• The description notes the proposed method has increased computational cost and memory usage for large-scale graphs.
• Forecasting performance is influenced by graph design and edge feature quality, which may not be fully documented in the data.

Provenance

Source

figshare, author Mengze Gu.

Collection Method

Likely contains experimental results from an ablation study comparing forecasting models.

Time Range

null

Freshness

Last updated 2026-04-23 17:25:32; freshness should be verified.

Geography

null