TY - GEN
T1 - Statistical methods for longitudinal data with agricultural applications
AU - Anantharama Ankinakatte, Smitha
PY - 2014/5/13
Y1 - 2014/5/13
N2 - The PhD study focuses on modeling two kings of longitudinal data arising in agricultural applications: continuous time series data and discrete longitudinal data. Firstly, two statistical methods, neural networks and generalized additive models, are applied to predict masistis using multivariate continuous time series data obtained from robotic milking of cows, and their performance is compared. Secondly, the use of acyclic probabilistic finite automata /APFA) to model univariate discret longitudinal data is studied from a statistical modeling perspective, leading to a modified model selection algorithm. This was found to compare favourably with the algorithm implemented in the well-known Beagle software. Finally, an R package to apply APFA models developed as part of the PhD project is described
AB - The PhD study focuses on modeling two kings of longitudinal data arising in agricultural applications: continuous time series data and discrete longitudinal data. Firstly, two statistical methods, neural networks and generalized additive models, are applied to predict masistis using multivariate continuous time series data obtained from robotic milking of cows, and their performance is compared. Secondly, the use of acyclic probabilistic finite automata /APFA) to model univariate discret longitudinal data is studied from a statistical modeling perspective, leading to a modified model selection algorithm. This was found to compare favourably with the algorithm implemented in the well-known Beagle software. Finally, an R package to apply APFA models developed as part of the PhD project is described
M3 - PhD thesis
SN - 978-87-93176-09-6
PB - Aarhus University, Faculty of Science and Technology
ER -