Improving the efficiency of modelling longitudinal physical activity using compositional data analysis

Compositional data analysis has emerged in recent years to analyse physical activity data, including objective measurements from accelerometers. Compositional data analysis accounts for the dependence of physical activity behaviours in a single model, compared to standard statistical techniques, which incorrectly assumes behaviours are independent. This is done by analysing physical activity as proportions, rather than analysing the number of minutes spent in each physical activity behaviour. There has been limited progress in extending compositional data analysis over multiple time points (i.e. longitudinally). The methods developed as part of this PhD give a more informative picture of the association between physical activity and body weight. Although these methods have been demonstrated using physical activity data, they can be applied to any data that is compositional therefore making it applicable to many data sets.