Mokuzai Gakkaishi
Online ISSN : 1880-7577
Print ISSN : 0021-4795
ISSN-L : 0021-4795
CategoryI
Non-Normal Probability Ellipse for Representing Bivariate Data of the Strength Properties of Timber I.
Quantile-based non-normal probability ellipse
Satomi Sonoda Kazuhiko FujitaYasuo Iijima
Author information
JOURNAL FREE ACCESS

2014 Volume 60 Issue 2 Pages 80-86

Details
Abstract
Utilization of joint stochastic nature of timber strength properties such as elasticity and bending strength often produces good results. For example, the mechanical grading method provided an advanced system of allowable stresses. Statistical and visual expression methods are necessary for analyzing bivariate non-normal distributed data such as bivariate strength properties of timber. A probability ellipse is a graphical method for representing a bivariate distribution with specified inner probability. However it is not suitable for non-normal distributions. This report presents a method for drawing a skewed probability ellipse with marginal non-normal distributions. This method converts a conventional probability ellipse into a quantile-based non-normal ellipse using iso-probabilistic transformations based on marginal distributions. We have verified the following performances of the method by numerical experiments using some groups of accumulated data of tests in bending. The performances included accuracy of inner probability, uniformity of probability for de-correlated vectorial angle, and comparability among some data groups. These results confirmed that the method showed good performances.
Content from these authors
© 2014 by The Japan Wood Research Society
Previous article Next article
feedback
Top