Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.20392497 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866902102943465472 |
|---|---|
| author | Daniel agyekum Amakye |
| author_facet | Daniel agyekum Amakye |
| contents | <h2>First Release</h2> <p>This is the first official release of the Heteroscedastic Gaussian Process Regression (HGPR) model for soil pH prediction in reclaimed mine lands.</p> <h3>Key Results</h3> <ul> <li>R² = 0.886 predictive accuracy</li> <li>10.9% improvement in prediction interval calibration over standard GPR</li> <li>94.6% coverage in acidic zones (pH < 5.0) vs 86.1% for standard GPR</li> </ul> <h3>Model Features</h3> <ul> <li>Two-stage heteroscedastic GPR with residual variance modeling</li> <li>Standard GPR baseline for comparison</li> <li>Synthetic data generation with controlled heteroscedastic noise</li> <li>Stratified spatial train-test split</li> <li>Comprehensive evaluation metrics (RMSE, R², coverage probability, interval width)</li> </ul> <h3>Repository Structure</h3> <ul> <li><code>main_balanced.py</code> - Main working script</li> <li><code>data/</code> - Data generation and preprocessing</li> <li><code>models/</code> - GPR and HGPR implementations</li> <li><code>evaluation/</code> - Metrics and cross-validation</li> <li><code>visualization/</code> - Plotting utilities</li> </ul> <h3>Results Files</h3> <ul> <li><code>hgpr_results_balanced.csv</code> - Detailed predictions for 101 test samples</li> <li><code>hgpr_results_balanced.png</code> - Publication-ready visualization</li> <li><code>model_comparison_balanced.csv</code> - Summary comparison table</li> </ul> <h3>How to Run</h3> <pre><code class="language-bash">pip install -r requirements.txt python main_balanced.py </code></pre> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_20392497 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | DanielAgyekumAmakyeGh/hgpr-mine-reclamation: Heteroscedastic GPR for Mine Reclamation pH Prediction - Version 1.0 Daniel agyekum Amakye <h2>First Release</h2> <p>This is the first official release of the Heteroscedastic Gaussian Process Regression (HGPR) model for soil pH prediction in reclaimed mine lands.</p> <h3>Key Results</h3> <ul> <li>R² = 0.886 predictive accuracy</li> <li>10.9% improvement in prediction interval calibration over standard GPR</li> <li>94.6% coverage in acidic zones (pH < 5.0) vs 86.1% for standard GPR</li> </ul> <h3>Model Features</h3> <ul> <li>Two-stage heteroscedastic GPR with residual variance modeling</li> <li>Standard GPR baseline for comparison</li> <li>Synthetic data generation with controlled heteroscedastic noise</li> <li>Stratified spatial train-test split</li> <li>Comprehensive evaluation metrics (RMSE, R², coverage probability, interval width)</li> </ul> <h3>Repository Structure</h3> <ul> <li><code>main_balanced.py</code> - Main working script</li> <li><code>data/</code> - Data generation and preprocessing</li> <li><code>models/</code> - GPR and HGPR implementations</li> <li><code>evaluation/</code> - Metrics and cross-validation</li> <li><code>visualization/</code> - Plotting utilities</li> </ul> <h3>Results Files</h3> <ul> <li><code>hgpr_results_balanced.csv</code> - Detailed predictions for 101 test samples</li> <li><code>hgpr_results_balanced.png</code> - Publication-ready visualization</li> <li><code>model_comparison_balanced.csv</code> - Summary comparison table</li> </ul> <h3>How to Run</h3> <pre><code class="language-bash">pip install -r requirements.txt python main_balanced.py </code></pre> |
| title | DanielAgyekumAmakyeGh/hgpr-mine-reclamation: Heteroscedastic GPR for Mine Reclamation pH Prediction - Version 1.0 |
| url | https://doi.org/10.5281/zenodo.20392497 |