The data contains an outlying cluster of observations 33 to 38 a second outlier
cluster of observations 7 to 11 and a few more isolated outliers, namely observations
12, 13, 31 and 32.
bushfirem is created from bushfire by setting a proportion of 0.2 of the values
to missing.
For testing purposes weights are provided:
bushfire.weights <- rep(c(1,2,5), length = nrow(bushfire))