12.08.2013 Views

final_program_abstracts[1]

final_program_abstracts[1]

final_program_abstracts[1]

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

11 IMSC Session Program<br />

The metaverification minefield<br />

Wednesday - Parallel Session 11<br />

Ian Jolliffe and David Stephenson<br />

University of Exeter, UK<br />

A common approach in forecast verification is to construct a numerical score that<br />

measures the relative quality of different forecasts. For a given type of forecast data<br />

there are typically a number of possible scores or measures that can be used. A<br />

question that then arises is how to choose between candidate scores. For this purpose<br />

a number of properties of scores were defined in the 1980s and 1990s, by Allan<br />

Murphy and others, which it is deemed that ‘good’ scores should possess. This<br />

exercise of looking at properties of scores is known as metaverification.<br />

It has become increasingly apparent that metaverification is a complex topic, with<br />

properties such as propriety, equitability and consistency having a number of hidden<br />

subtleties. Some of these are explored in this presentation, including the<br />

incompatibility of propriety and equitability, and the non-equitability of the ‘equitable<br />

threat score’.<br />

Abstracts 190

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!