{"title":"A method to improve binary forecast skill verification","authors":"Thitithep Sitthiyot , Kanyarat Holasut","doi":"10.1016/j.mex.2024.103010","DOIUrl":null,"url":null,"abstract":"<div><div>To overcome the limitations of existing deterministic binary forecast skill verification methods that award a perfect score for forecasting events considered easy to forecast, an improvement factor is introduced. It comprises two components which are 1) a measure of the ease with which an event can be accurately forecasted and 2) a measure of frequency of event. By using two hypothetical datasets, this study demonstrates that an improvement factor could enhance the performance of existing deterministic binary forecast skill verification methods by awarding score that is close to score for no-skill forecast for the perfect forecasts of events considered easy to forecast. In addition, the forecast and actual data on annual inflation rate are used to demonstrate how an improvement factor could be used together with the existing deterministic binary forecast skill verification methods in order to assess skills of the forecasters in practice.<ul><li><span>•</span><span><div>Existing deterministic binary forecast skill verification methods fail to award correct score for events considered easy to forecast.</div></span></li><li><span>•</span><span><div>An improvement factor is developed in order to enhance performance of existing deterministic binary forecast skill verification methods.</div></span></li><li><span>•</span><span><div>Hypothetical and empirical data are used to validate how an improvement factor works.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"13 ","pages":"Article 103010"},"PeriodicalIF":1.6000,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11513477/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"MethodsX","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2215016124004618","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MULTIDISCIPLINARY SCIENCES","Score":null,"Total":0}
引用次数: 0
Abstract
To overcome the limitations of existing deterministic binary forecast skill verification methods that award a perfect score for forecasting events considered easy to forecast, an improvement factor is introduced. It comprises two components which are 1) a measure of the ease with which an event can be accurately forecasted and 2) a measure of frequency of event. By using two hypothetical datasets, this study demonstrates that an improvement factor could enhance the performance of existing deterministic binary forecast skill verification methods by awarding score that is close to score for no-skill forecast for the perfect forecasts of events considered easy to forecast. In addition, the forecast and actual data on annual inflation rate are used to demonstrate how an improvement factor could be used together with the existing deterministic binary forecast skill verification methods in order to assess skills of the forecasters in practice.
•
Existing deterministic binary forecast skill verification methods fail to award correct score for events considered easy to forecast.
•
An improvement factor is developed in order to enhance performance of existing deterministic binary forecast skill verification methods.
•
Hypothetical and empirical data are used to validate how an improvement factor works.