A bivariate model for retinal image identification in lambs

In this paper we introduce the one-inflated bivariate Beta distribution as a new approach to analyze matching scores related to retinal image identification in lambs. These kind of data sets present many values equal to one, the perfect matching, and this justifies the introduction of one-inflation patterns. In addition, matching scores are correlated according to several bivariate patterns: left and right eye, 3 and 12 months of age, etc. Some examples of applications are given.

► New statistical method to analyses the matching scores of images obtained with Optibrand device.
► This method takes into account the continuous and semi-continuous data.
► R programs, available from the authors, are used to implement these models.