Hello, From Tang et.al in An analysis of diversity measures (DOI 10.1007/s10994-006-9449-2) stated that:
(1) diversity is greater when the measure is larger, such as: disagreement measure, Kohavi Wolpert, Generalized diversity.
(2) diversity is greater when the measure is smaller, such as: double fault, interrater aggreement, measure difficulty.
what I want to ask is about Q-statistic, Correlation coefficient and conditional double fault. where are those 3 belongs? in diversity is greater when the measure is larger or in diversity is greater when the measure is smaller?
thank you
I found the detailed explanation, comes from paper:
titles: "Creating and measuring diversity in multiple classifier systems using support vector data description".
Author: Haghighi et al
doi:10.1016/j.asoc.2011.06.006
title: "Measures of Diversity in Classifier Ensembles and Their Relationship with the Ensemble Accuracy"
Author: Kuncheva & Whitaker
doi: 10.1023/a:1022859003006
- Badges
- Science topic
- Similar topics
- Psychometrics
- Measurement