Random censoring and dendritic trees

The motivating problem is the estimation of the branching parameters of dendritic trees when some of the branches are cut. A primary element of this problem is the estimation of a bivariate discrete distribution when, because of partial censoring, some of the observations are incomplete. (This is similar to the problem of a multinomial cross-classification with supplemental partially cross-classified data.) Tests for randomness of censoring and estimators of parameters under random censoring are derived. The results on random censoring are applied in the dendritic tree context to obtain estimators for certain branching and cutting parameters. The testing and estimation procedures are applied to a set of dendritic tree data.