Datasets:

Dedulek
/

gaia-eb-teff-datasets

@@ -1,313 +1,146 @@
 ---
-title: Gaia Eclipsing Binary Teff Datasets
-language:
-  - en
 license: cc-by-4.0
-size_categories:
-  - 1M<n<10M
 task_categories:
-  - tabular-regression
 tags:
-  - astronomy
-  - astrophysics
-  - eclipsing-binaries
-  - stellar-parameters
-  - photometry
-  - gaia
-  - effective-temperature
-pretty_name: Gaia Eclipsing Binary Effective Temperature Datasets
 ---
-# Gaia Eclipsing Binary Effective Temperature Datasets
-## Dataset Description
-This dataset contains multi-survey photometry and stellar parameters for **2.18 million eclipsing binary stars** from the Gaia mission. It combines data from Gaia DR3, Pan-STARRS DR1, and 2MASS to enable machine learning prediction of effective temperatures (Teff) for stars lacking spectroscopic measurements.
-### Dataset Summary
-- **Total objects**: 2,179,680 eclipsing binary stars
-- **Gaia DR3 coverage**: 100% (all sources have Gaia photometry)
-- **Pan-STARRS coverage**: 53.5% (1,166,000 sources)
-- **2MASS coverage**: Variable (J, H, K bands)
-- **Teff coverage**: 58% have Gaia GSP-Phot temperatures
-- **ML predictions**: 38.9% (847,000 stars) have ML-predicted temperatures
-### Surveys Included
-1. **Gaia DR3** (2023)
-   - G, BP, RP magnitudes and colors
-   - GSP-Phot effective temperatures
-   - Astrometric parameters
-2. **Pan-STARRS DR1** (2016)
-   - g, r, i, z, y optical magnitudes
-   - PSF and Kron photometry
-3. **2MASS** (2003)
-   - J, H, K near-infrared magnitudes
-## Dataset Structure
-### Unified Photometry Dataset
-**File**: `photometry/eb_unified_photometry.parquet`
-**Size**: 227 MB
-**Format**: Apache Parquet
-This is the primary dataset containing all photometry and stellar parameters.
-#### Key Columns
-**Identifiers:**
-- `source_id` (int64): Gaia DR3 source identifier
-**Gaia Photometry:**
-- `g`, `bp`, `rp` (float64): Gaia G, BP, RP magnitudes
-- `bp_rp`, `g_bp`, `g_rp` (float64): Gaia colors
-- `parallax`, `pmra`, `pmdec` (float64): Astrometry
-**Gaia Stellar Parameters:**
-- `teff_gaia` (float64): GSP-Phot effective temperature [K]
-- `logg_gaia` (float64): Surface gravity [log cm/s²]
-- `mh_gaia` (float64): Metallicity [Fe/H]
-**Pan-STARRS Photometry:**
-- `ps_gPSFMag`, `ps_rPSFMag`, `ps_iPSFMag`, `ps_zPSFMag`, `ps_yPSFMag` (float64): PSF magnitudes
-- `ps_gKronMag`, `ps_rKronMag`, etc. (float64): Kron magnitudes
-- Pan-STARRS colors: `ps_g_r`, `ps_r_i`, etc.
-**2MASS Photometry:**
-- `j_m`, `h_m`, `k_m` (float64): 2MASS magnitudes
-- `j_h`, `h_k`, `j_k` (float64): 2MASS colors
-**Missing Values:**
-All missing values are encoded as `-999.0` for consistency.
-### Final Catalog with Predictions
-**File**: `catalogs/stars_types_with_best_predictions.fits`
-**Size**: 196 MB
-**Format**: FITS binary table
-Complete catalog of 2.1M eclipsing binaries with:
-- Original Gaia temperatures (where available)
-- ML-predicted temperatures (best-of-three ensemble)
-- Prediction uncertainties
-- Quality flags (A=Gaia, B/C/D=ML by uncertainty, X=none)
-**Coverage**: 97.2% of stars have Teff values (58.3% Gaia original + 38.9% ML predictions)
-## Usage
-### Download with Python
 ```python
-from huggingface_hub import hf_hub_download
-import polars as pl
-# Download unified photometry
-file_path = hf_hub_download(
-    repo_id="Dedulek/gaia-eb-teff-datasets",
-    filename="photometry/eb_unified_photometry.parquet",
-    repo_type="dataset"
-)
-# Load with Polars (recommended for large datasets)
-df = pl.read_parquet(file_path)
-# Or with Pandas
 import pandas as pd
-df = pd.read_parquet(file_path)
-print(f"Loaded {len(df)} eclipsing binaries")
-print(f"Columns: {df.columns}")
-```
-### Download with Hugging Face CLI
-```bash
-# Install CLI
-pip install huggingface_hub
-# Download specific file
-huggingface-cli download Dedulek/gaia-eb-teff-datasets \
-    --repo-type dataset \
-    --include "photometry/eb_unified_photometry.parquet" \
-    --local-dir ./data
-```
-### Training Example
-```python
-import polars as pl
-from sklearn.ensemble import RandomForestRegressor
-from sklearn.model_selection import train_test_split
-# Load data
-df = pl.read_parquet("eb_unified_photometry.parquet")
-# Filter stars with known Teff and Gaia photometry
-df_train = df.filter(
-    (pl.col("teff_gaia") != -999.0) &
-    (pl.col("bp_rp") != -999.0)
-)
-# Prepare features and target
-features = ["g", "bp", "rp", "bp_rp", "g_bp", "g_rp"]
-X = df_train[features].to_numpy()
-y = df_train["teff_gaia"].to_numpy()
-# Train model
-X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
-model = RandomForestRegressor(n_estimators=300, max_depth=20, random_state=42)
-model.fit(X_train, y_train)
-# Evaluate
-score = model.score(X_test, y_test)
-print(f"R² score: {score:.3f}")
-```
-## Dataset Statistics
-### Photometric Coverage
-| Survey | Coverage | N_stars |
-|--------|----------|---------|
-| Gaia DR3 | 100% | 2,179,680 |
-| Pan-STARRS DR1 | 53.5% | 1,166,000 |
-| 2MASS (J) | ~60% | ~1,300,000 |
-| 2MASS (H) | ~60% | ~1,300,000 |
-| 2MASS (K) | ~60% | ~1,300,000 |
-### Stellar Parameter Coverage
-| Parameter | Coverage | Mean | Std | Range |
-|-----------|----------|------|-----|-------|
-| Teff (Gaia) | 58% | 7,450 K | 3,200 K | 2,500 - 50,000 K |
-| log(g) | 56% | 3.8 | 0.5 | 0.5 - 5.5 |
-| [Fe/H] | 48% | -0.2 | 0.4 | -2.5 - +0.5 |
-### Temperature Distribution
-| Teff Range | N_stars | Percentage |
-|------------|---------|------------|
-| < 4,000 K (Cool) | 180,000 | 14% |
-| 4,000-6,000 K (Mid) | 520,000 | 41% |
-| 6,000-10,000 K (Hot) | 450,000 | 36% |
-| > 10,000 K (Very Hot) | 115,000 | 9% |
-## Data Quality
-### Missing Value Convention
-All surveys use `-999.0` to indicate missing values. Always filter these before analysis:
-```python
-# Filter valid measurements
-df_clean = df.filter(
-    (pl.col("bp_rp") != -999.0) &
-    (pl.col("teff_gaia") != -999.0)
-)
 ```
-### Known Issues
-1. **Gaia GSP-Phot Bias**: Systematic underestimation of Teff for hot stars (>10,000 K)
-   - Correction coefficients available in model repository
-   - See: `data/teff_correction_coeffs_deg2.pkl`
-2. **Pan-STARRS Coverage**: Northern hemisphere bias (Dec > -30°)
-3. **2MASS Saturation**: Bright stars (J < 6) may be saturated
-## Model Performance
-Pre-trained models achieve the following performance on held-out test sets:
-| Model | Features | MAE (K) | RMSE (K) | R² | Within 10% |
-|-------|----------|---------|----------|-----|------------|
-| Gaia Colors (Log) | 6 Gaia colors/bands | 557 | 1,021 | 0.640 | 68.5% |
-| Gaia + 2MASS | 5 optical+IR colors | 765 | 1,168 | 0.315 | 43.4% |
-| Best-of-Three Ensemble | Multiple models | 263 | - | - | - |
 ## Citation
 If you use this dataset, please cite:
 ```bibtex
-@dataset{gaia_eb_teff_2025,
-  author = {Your Name},
-  title = {Gaia Eclipsing Binary Effective Temperature Datasets},
-  year = {2025},
-  publisher = {HuggingFace},
-  url = {https://huggingface.co/datasets/Dedulek/gaia-eb-teff-datasets}
 }
 ```
-### Data Sources
-Please also cite the original surveys:
-**Gaia DR3:**
-```bibtex
-@article{gaia2023,
-  author = {{Gaia Collaboration}},
-  title = {Gaia Data Release 3},
-  journal = {Astronomy & Astrophysics},
-  year = {2023},
-  volume = {674},
-  pages = {A1}
-}
-```
-**Pan-STARRS:**
-```bibtex
-@article{panstarrs2020,
-  author = {Flewelling, H. A. and others},
-  title = {The Pan-STARRS1 Database and Data Products},
-  journal = {The Astrophysical Journal Supplement Series},
-  year = {2020},
-  volume = {251},
-  pages = {7}
-}
-```
-**2MASS:**
-```bibtex
-@article{2mass2006,
-  author = {Skrutskie, M. F. and others},
-  title = {The Two Micron All Sky Survey (2MASS)},
-  journal = {The Astronomical Journal},
-  year = {2006},
-  volume = {131},
-  pages = {1163}
-}
-```
 ## License
-This dataset is released under **CC BY 4.0** (Creative Commons Attribution 4.0 International).
-You are free to:
-- Share: copy and redistribute the material
-- Adapt: remix, transform, and build upon the material
-Under the following terms:
-- Attribution: You must give appropriate credit and indicate if changes were made
-## Contact
-For questions or issues with this dataset:
-- Open an issue on the [GitHub repository](https://github.com/YOUR_USERNAME/gaia-eb-teff)
-- Contact: your.email@example.com
-## Acknowledgments
-This work has made use of data from:
-- ESA mission Gaia (https://www.cosmos.esa.int/gaia)
-- Pan-STARRS (https://panstarrs.stsci.edu/)
-- 2MASS (https://www.ipac.caltech.edu/2mass/)
-Special thanks to the astronomical community for making these datasets publicly available.

 ---
 license: cc-by-4.0
 task_categories:
+- tabular-regression
 tags:
+- astronomy
+- astrophysics
+- eclipsing-binaries
+- gaia
+- stellar-parameters
+- temperature-prediction
+size_categories:
+- 1M<n<10M
 ---
+# Gaia Eclipsing Binary Effective Temperature Dataset
+This dataset provides effective temperature (Teff) predictions for 2.1 million eclipsing binary stars using a best-of-four ensemble approach combining Random Forest models trained on Gaia DR3 photometry.
+## Dataset Summary
+**Total Objects**: 2,145,310 eclipsing binaries
+**Teff Coverage**: 97.2% (2,085,712 objects with temperature estimates)
+**High Quality (A+B flags)**: 92.7% (1,933,634 objects)
+**Data Sources**: Gaia DR3, Pan-STARRS, 2MASS
+## Best-of-Four Ensemble
+The dataset uses a best-of-four ensemble that selects the prediction with the **lowest Random Forest uncertainty** for each object from four models:
+1. **Teff-only model** (colors only, corrected Gaia temperatures)
+2. **Teff+logg model** (colors + surface gravity)
+3. **Teff+clustering model** (colors + cluster features)
+4. **Flag 1 model** (trained on highest quality Gaia sources)
+**Performance**:
+- Mean uncertainty: **203 K** (22.8% improvement over single best model)
+- Model selection: 29.2% flag 1, 25.6% colors-only, 23.8% +logg, 21.4% +clustering
+## Files
+### Main Catalog
+- **`catalogs/stars_types_with_best4_predictions.fits`** (206 MB)
+  - Complete catalog with best-of-four Teff predictions
+  - Columns: `source_id`, `ra`, `dec`, `teff_final`, `teff_uncertainty`, `quality_flag`, `best_model`
+  - See accompanying `_DESCRIPTION.txt` for full schema
+### Photometry
+- **`photometry/eb_unified_photometry.parquet`** (69 MB)
+  - Gaia DR3 + Pan-STARRS + 2MASS photometry for all eclipsing binaries
+  - Includes magnitudes, colors, and photometric uncertainties
+### Correction Coefficients
+- **`correction/teff_correction_coeffs_deg2.pkl`**
+  - Polynomial coefficients for correcting Gaia GSP-Phot temperatures (>10,000K)
+  - Used during model training to correct systematic underestimation
+## Quality Flags
+| Flag | Description | Uncertainty | Count | Percentage |
+|------|-------------|-------------|-------|------------|
+| **A** | Gaia GSP-Phot original | N/A | 1,249,946 | 58.3% |
+| **B** | ML prediction (< 300K) | < 300K | 683,688 | 31.9% |
+| **C** | ML prediction (300-500K) | 300-500K | 152,078 | 7.1% |
+| **D** | ML prediction (≥ 500K) | ≥ 500K | 6,000 | 0.3% |
+| **X** | No temperature available | N/A | 53,598 | 2.5% |
+**High Quality (A+B)**: 92.7% of catalog
+**ML Predictions (B+C+D)**: 39.2% (841,766 objects)
+## Usage Example
 ```python
+from astropy.table import Table
 import pandas as pd
+# Load FITS catalog
+catalog = Table.read('stars_types_with_best4_predictions.fits')
+# Convert to pandas for analysis
+df = catalog.to_pandas()
+# Filter high-quality predictions
+high_quality = df[df['quality_flag'].isin(['A', 'B'])]
+# Get objects with low uncertainty ML predictions
+low_unc_ml = df[(df['quality_flag'] == 'B') & (df['teff_uncertainty'] < 200)]
+print(f"High quality sources: {len(high_quality):,}")
+print(f"Low uncertainty ML predictions: {len(low_unc_ml):,}")
 ```
+## Model Information
+Pre-trained models are available in the companion repository:
+**[Dedulek/gaia-eb-teff-models](https://huggingface.co/Dedulek/gaia-eb-teff-models)**
+### Flag 1 Model (Primary contributor to ensemble)
+- **Model**: `rf_gaia_colors_flag1_20251218_094937.pkl`
+- **Training**: 711,666 high-quality Gaia sources (flag 1)
+- **Features**: 6 Gaia colors + bands (g, bp, rp, bp_rp, g_rp, g_bp)
+- **Performance**: MAE 195K, R² 0.9126
+- **Contribution**: Selected for 29.2% of ML predictions
 ## Citation
 If you use this dataset, please cite:
 ```bibtex
+@dataset{gaia_eb_teff_2024,
+  title={Gaia Eclipsing Binary Effective Temperature Dataset},
+  author={Your Name/Team},
+  year={2024},
+  publisher={Hugging Face},
+  howpublished={\url{https://huggingface.co/datasets/Dedulek/gaia-eb-teff-datasets}}
 }
 ```
+## Data Processing
+This dataset was created using a configurable ML pipeline:
+1. **Training Data**: Gaia DR3 GSP-Phot temperatures with polynomial correction (>10,000K)
+2. **Features**: Gaia photometry (g, bp, rp bands) + derived colors
+3. **Model**: Random Forest with full tree uncertainty estimation
+4. **Ensemble**: Best-of-four selection based on RF uncertainty
+5. **Quality Control**: Systematic validation against spectroscopic surveys
 ## License
+This dataset is released under **CC BY 4.0**. You are free to share and adapt with attribution.
+## Related Resources
+- **Models**: [Dedulek/gaia-eb-teff-models](https://huggingface.co/Dedulek/gaia-eb-teff-models)
+- **Gaia Archive**: [https://gea.esac.esa.int/archive/](https://gea.esac.esa.int/archive/)
+- **Pan-STARRS**: [https://panstarrs.stsci.edu/](https://panstarrs.stsci.edu/)
+## Updates
+- **2024-12-18**: Updated to best-of-four ensemble (mean uncertainty: 203K)
+- Previous: best-of-three ensemble (mean uncertainty: 263K)
+## Contact
+For questions or issues, please open an issue in the dataset repository or contact the authors.