全部 标题 作者
关键词 摘要


Model Selection by Friedman Statistics

DOI: 10.1234/pjsor.v7i2-sp.285

Keywords: Model Selection , Econometric Model , Friedman Statistic , Prediction Performance , Post Estimation Data

Full-Text   Cite this paper   Add to My Lib

Abstract:

This study investigates an application of Friedman statistic as a model selection methodology on post estimation data. The Friedman statistic is employed for testing the possible differences between related samples by ranking the data. Similarly, we suggest ranking of competing models based on a specific multiple comparison procedure for identifying differences between models. As a non-parametric test statistic, it does not make assumptions regarding the underlying distribution of data, however, this procedure may require a large dataset since it relies on post estimation comparisons.

Full-Text

comments powered by Disqus