Case 1 Ablation Study Results for Graph-Based Load Forecasting

Ablation study results comparing an Edge Graph Attention Network (EGAT) model against GRU, MLP, and LSTM methods for power load forecasting. The 5.5 KB XLS file, authored by Mengze Gu and last updated in April 2026, contains results from a study that transforms time series into graph features. The study demonstrates EGAT's effectiveness in finding important features and understanding complex time patterns for energy demand prediction.

Use Cases

• Benchmarking graph-based forecasting models against traditional methods like GRU and LSTM based on the described comparison.
• Analyzing the impact of graph construction on forecasting accuracy based on the described connectivity patterns and edge features.
• Evaluating computational cost trade-offs for attention-based graph models based on the mentioned training time and memory usage considerations.

Strengths

• Results are from a direct comparison of four distinct model architectures (EGAT, GRU, MLP, LSTM).
• Dataset is openly licensed under CC-BY-4.0, permitting reuse with attribution.
• File is a small 5.5 KB XLS, facilitating quick inspection and analysis.

Limitations

• Row count and specific column-level documentation are absent; field semantics must be inferred after download.
• The dataset's scale is limited to a single ablation study ('Case 1'), which may not represent broader performance.
• Freshness should be verified as the last update date is in the future (2026-04 23).

Provenance

Source

figshare, author Mengze Gu.

Collection Method

Likely contains numerical results from a model comparison experiment described in the associated paper.

Time Range

null

Freshness

Last updated 2026-04-23 17:25:30; freshness should be verified.

Geography

null