Analysis of faeces particles size distribution from ruminating species using linear modelling in the Aitchison geometry

Specialeforsvar: Simona Galkova

Titel: Analysis of faeces particles size distribution from ruminating species using linear modelling in the Aitchison geometry

Resume: An applied statistical analysis of the faces particles sizes data from ruminating species is considered in this thesis. Log-probability density functions are studied in Hilbert space using the properties of Aitchison geometry. On this Hilbert space, a finite Fourier
expansion of the log-probability density functions is made, where the Fourier coefficients serve as an observed data for modelling. Using the multivariate statistics, a linear normal model for the log-probability density functions considering both the random and the fixed effect is formulated. Parameter estimates from the model are found using the EM-algorithm. It is investigated if this approach fits the data based on standard validation techniques. Finally, further biological hypotheses are investigated and the results are
interpreted in terms of associated additive (multiplicative or power-related, respectively) effect on the probability density functions in Aitchison geometry

Vejledere: Bo Markussen og Peder Nørgaard
Censor: Bo Martin Bibby