Benchmark连续排名概率得分 / Continuous Ranked Probability Score

CRPS

A metric for evaluating the quality of probabilistic forecasts that measures how closely a predictive distribution matches the true observation, accounting for both accuracy and calibration.

Timeline (last 90 days)

Nov 5

CRPS measures the integral of squared error between the predicted cumulative distribution function and the true observed value, penalizing both overconfidence and underconfidence.

Unverified50%

Nov 4

Among the 17 datasets in the ProbTS benchmark suite, KGO achieves the best CRPS on 13 datasets and the best NMAE on 16 datasets.

Unverified50%

All Facts (2)

Unverified

CRPS measures the integral of squared error between the predicted cumulative distribution function and the true observed value, penalizing both overconfidence and underconfidence.

50%Unverified

Among the 17 datasets in the ProbTS benchmark suite, KGO achieves the best CRPS on 13 datasets and the best NMAE on 16 datasets.

50%

Source Articles

FlowReconBench: Sparse-Sensor Flow Field Reconstruction — Are Learned Models or Classical Methods Stronger?Nov 5 Koopman Generative Operators: Single-Step Inference for Efficient Probabilistic Time Series ForecastingNov 4