A comprehensive analysis pipeline for quantifying how tree cover affects electricity consumption under different temperature conditions across China (2012-2019).
This project analyzes 11.65 million grid-month observations to understand how urban tree cover can reduce energy consumption during heat waves. Using advanced econometric methods, we quantify the cooling benefits of trees and identify where green infrastructure investments would be most effective.
- Processes massive climate and energy datasets (>100GB) at 1km resolution
- Quantifies tree cooling benefits with precise economic metrics
- Identifies optimal locations for urban forestry investments
- Tests climate conditions where trees are most/least effective
- Provides reproducible analysis pipeline for other regions/countries
- 33-48% reduction in temperature-induced electricity consumption in tree-covered areas
- Peak effectiveness at 30-35°C (48% cooling benefit)
- Reduced effectiveness under high humidity conditions (transpiration failure)
- Quantified energy savings for cost-benefit analysis of green infrastructure
- Python 3.8+: Data processing, ML experiments
- R 4.0+: Econometric analysis, visualization
- Git: Version control
- Python: pandas, numpy, scikit-learn, lightgbm, xgboost, shap
- R: fixest, ggplot2, sf, raster, dplyr
- Spatial Resolution: 1km × 1km grid cells
- Temporal Coverage: Monthly data from 2012-2019
- Data Volume: >100GB raw data processed
code-test/
├── scripts/ # Analysis pipeline
│ ├── Download-Cleaning/ # Data acquisition scripts
│ ├── Data-Processing/ # Data integration & processing
│ └── ML-Analysis/ # Machine learning experiments
├── config/ # Configuration files
│ ├── download-config.json # API keys and download settings
│ ├── data-processing.json # Processing parameters
│ └── ml-analysis.json # ML model configurations
├── data/ # Data storage (contents gitignored)
│ ├── raw/ # Original downloaded datasets
│ ├── processed/ # Processed analysis-ready data
│ └── shp/ # Shapefiles for geographic boundaries
├── r-analysis/ # R econometric analysis
│ ├── geo-analysis.Rmd # Main analysis notebook
│ ├── results/ # Regression output files
│ └── figs/ # Generated figures
├── output/ # Analysis outputs (gitignored)
│ ├── feature_selection/ # Feature selection results
│ ├── ml_pipeline/ # ML model results
│ └── shap_analysis/ # SHAP interpretability results
├── logs/ # Processing logs (gitignored)
├── requirements.txt # Python dependencies
└── README.md # Project documentation
Note: The data/, output/, and logs/ directories contain large files that are not included in the repository. You'll need to generate these by running the data download and processing scripts.
# Python 3.8+
pip install -r requirements.txt
# R 4.0+ with required packages
install.packages(c("fixest", "ggplot2", "sf", "raster", "dplyr"))git clone https://github.com/stefanfux0115/code-test.git
cd code-testEdit the JSON files in config/ directory to set your local data paths and API keys.
# Download raw data (requires API keys)
python scripts/Download-Cleaning/download_master.py --all
# Process and integrate data
python scripts/Data-Processing/unified_data_processor_optimized.py --all
# Run econometric analysis
# Open r-analysis/geo-analysis.Rmd in RStudioThis project is licensed under the MIT License - see LICENSE file.
Enxian Fu
Email: stefanfu015@gmail.com
- Research advisors and collaborators
- Data providers (ERA5, Chen et al., ChinaMet)
- Open source community
Star this repo if you find it useful!
Found a bug? Open an issue!
Have questions? Feel free to reach out!