Provincial Carbon Emission Forecasts for China 2021-2025

2021 to 2025 provincial carbon emission data supports a federated learning framework dividing China into five macro-regions. The study demonstrates forecasting improvements, with MAE reductions from 2.68% to 17.91% across regions.

Use Cases

• Validate federated learning models using the five macro-region partitioning strategy for spatiotemporal heterogeneity analysis.
• Benchmark interval forecasting and uncertainty quantification methods against reported PICP metrics exceeding 96%.
• Analyze regional performance differences using reported MAE reductions and R² improvements per macro-region.

Strengths

• Data underpins a published methodological study with quantified performance gains (e.g., 17.91% MAE reduction).
• Explicit temporal coverage from 2021 to 2025 for provincial-level analysis.
• Framework addresses spatial heterogeneity via a defined five macro-region structure (e.g., Eastern, Northwestern).

Limitations

• The core dataset is not directly accessible; analysis is based on a 1021.2 KB DOCX supplementary file describing results.
• Unknown row and column counts for the underlying provincial emission time series data.
• Geographic scope is limited to China's provinces.

Provenance

Source

figshare, authored by Yonggang Xiao.

Collection Method

Derived from provincial carbon emission data, analyzed via a personalized federated learning (pFL) framework with LSTM-ASA.

Time Range

2021 to 2025.

Freshness

null

Geography

China, partitioned into five macro-regions (e.g., Eastern, Northwestern).