properscoring: Proper scoring rules in Python. Proper scoring rules for evaluating probabilistic forecasts in Python. Evaluation methods that are “strictly proper” cannot be artificially improved through hedging, which makes them fair methods for accessing the accuracy of probabilistic forecasts. In particular, these rules are often used for evaluating weather forecasts. properscoring runs on both Python 2 and 3. It requires NumPy (1.8 or later) and SciPy (any recent version should be fine). Numba is optional, but highly encouraged: it enables significant speedups (e.g., 20x faster) for crps_ensemble and threshold_brier_score. To install, use pip: pip install properscoring.
Keywords for this software
References in zbMATH (referenced in 1 article )
Showing result 1 of 1.
- Alexander Jordan, Fabian Krueger, Sebastian Lerch: Evaluating probabilistic forecasts with the R package scoringRules (2017) arXiv