Home Page


When working with temporal data it is important to capture both long and short term trends. In this series I’m modelling the relationship between the weight of banded mongooses over the course of their lives. The purpose is to compare several the performance of several model at this task.

This time the focus is on Additive Mixed Modelling. The relationship with time is modelled using a smoother (in this case two smoothers, one for long and one for short term effects). We also fit a random intercept to account for differences between individuals. Finally we also fit constraints on the residuals to account for heterogeneity and auto-correlation.

Data Exploration

First lets look at our data.

xyplot(train$weight ~ train$agen/365,
       xlab="Age in years", ylab="Weight in g")