A distance based test on random trees

ANA GEORGINA FLESIA

Córdoba, Argentina

Congreso; MACI 2007 I CONGRESS ON COMPUTATIONAL, INDUSTRIAL AND APPLIED MATHEMATICS; 2007

FaMAF- UNC y Sociedad Argentina de Mecánica Computacional

In this talk, we address the question of comparison between populations of trees. We study an statistical test based on thedistance between empirical mean trees, as an analog of the two sample Z statistic for comparing means of two gaussian distributed populations. Despite its simplicity, we can report that the test is  quite powerful toseparate distributions with different means but it does notdistinguish between different populations with the same mean tree. In this case, we suggest to apply a more complicated test, the BFFS test, if the problem allows its computation.   The performance of the distance test is studied viasimulations on Galton-Watson branching processes. We also show an application  to a real data problem in genomics.