Using scoring functions to evaluate point process forecasts

03/22/2021

∙

Point process models are widely used tools to issue forecasts or assess risks. In order to check which models are useful in practice, they are examined by a variety of statistical methods. We transfer the concept of consistent scoring functions, which are principled statistical tools to compare forecasts, to the point process setting. The results provide a novel approach for the comparative assessment of forecasts and models and encompass some existing testing procedures.

READ FULL TEXT

Using scoring functions to evaluate point process forecasts

Of Quantiles and Expectiles: Consistent Scoring Functions, Choquet Representations, and Forecast Rankings

Order-Sensitivity and Equivariance of Scoring Functions

Evaluation of point forecasts for extreme events using consistent scoring functions

Proper Scoring Rules, Gradients, Divergences, and Entropies for Paths and Time Series

On Single Point Forecasts for Fat-Tailed Variables

Model Comparison and Calibration Assessment: User Guide for Consistent Scoring Functions in Machine Learning and Actuarial Practice

Long-Term Forecasts of Military Technologies for a 20-30 Year Horizon: An Empirical Assessment of Accuracy

Using scoring functions to evaluate point process forecasts

Related Research

Of Quantiles and Expectiles: Consistent Scoring Functions, Choquet Representations, and Forecast Rankings

Order-Sensitivity and Equivariance of Scoring Functions

Evaluation of point forecasts for extreme events using consistent scoring functions

Proper Scoring Rules, Gradients, Divergences, and Entropies for Paths and Time Series

On Single Point Forecasts for Fat-Tailed Variables

Model Comparison and Calibration Assessment: User Guide for Consistent Scoring Functions in Machine Learning and Actuarial Practice

Long-Term Forecasts of Military Technologies for a 20-30 Year Horizon: An Empirical Assessment of Accuracy