This figure shows how the proportion of households offline varies with median connection speed for each decile of (each 10% of the population ordered by) SIMD deprivation. Once we allow for the effects of deprivation, we can see that uptake generally increases where speeds are lower. In deciles 5-9 we see lower uptake where speeds are poor (< 4 Mb/s).
We use the model
dataset,
and fit the two parameters of a beta-binomial distribution, as smooth
functions of isolation, using a
generalised additive model, provided by the VGAM package in R. The lines are drawn using the following code:
with(model,
for (q in levels(decile)){
m <- subset(model,
decile == q & !is.na(medspeed))
speedgam <-
vgam(formula =
cbind(offline,
connected)
~ s(medspeed,df=4),
family = betabinomial.ab,
data = m,
trace = TRUE,
)
lines(test$medspeed,
predict(speedgam,
type="response",
newdata=test),
lwd=0.5)
symbols(m$medspeed,
predict(speedgam,type="response"),
circles=sqrt(m$Household_Count),
inches=0.1,
fg=alpha(ur6pal[m$UrbRur6_2011_2012],0.2),
bg=alpha(ur6pal[m$UrbRur6_2011_2012],0.2),
add=T)
text(32,
predict(speedgam,
type="response",
newdata=data.frame(medspeed=31)),
labels= c(q),adj=c(0,0.6))
}
)