These are Quantile Random Forest Regression models trained to predict the kpoints-density with different confidence levels:
- QRF95.pkl predicts (0.05, 0.5, 0.95) quantiles
- QRF90.pkl predicts (0.1, 0.5, 0.9) quantiles
- 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:
