|
|
--- |
|
|
license: cc-by-4.0 |
|
|
--- |
|
|
|
|
|
These are Quantile Random Forest Regression models trained to predict the kpoints-density with different confidence levels: |
|
|
1. QRF95.pkl predicts (0.05, 0.5, 0.95) quantiles |
|
|
2. QRF90.pkl predicts (0.1, 0.5, 0.9) quantiles |
|
|
3. QRF85.pkl predicts (0.15, 0.5, 0.85) quantiles |
|
|
|
|
|
The performance of models measured for the 0.5 quantile is: |
|
|
|
|
|
MAE: 0.064, |
|
|
|
|
|
MAPE: 0.179, |
|
|
|
|
|
MSE: 0.0098, |
|
|
|
|
|
R2_score: 0.694, |
|
|
|
|
|
Spearman_corr: 0.862, |
|
|
|
|
|
Kendall_corr: 0.682 |
|
|
|
|
|
Models are trained on the dataset generated for this work. |
|
|
|
|
|
Associated GitHub repositories: |
|
|
|
|
|
https://github.com/stfc/goldilocks |
|
|
|
|
|
https://github.com/stfc/goldilocks_kpoints |
|
|
|
|
|
|
