xyplot in R with several plots in one panel, lattice

Question

I have a test data set constructed in the following way:

d <- seq.Date(as.Date("2000-01-01"), as.Date("2000-01-08"), by=1)
df1 <- data.frame(time = d, type = 'type1', value = runif(length(d)))
df2 <- data.frame(time = d, type = 'group1', value = runif(length(d)))
df3 <- data.frame(time = d, type = 'group2', value = runif(length(d)))
df4 <- data.frame(time = d, type = 'pen', value = runif(length(d)))
df <- rbind(df1, df2, df3, df4)

I want to plot the data in lattice using the xyplot function, where the panels are displayed in one column. I can do the following:

xyplot(value~time | type, df, layout=c(1, length(levels(df$type))))

This way I have 4 rows (as many types I have). But, actually I would like to plot group1 and group2 in the same panel with different colors (so that I have only 3 rows). Could someone please help me out on this?

Answer 1

You can create one panel for all 'group' records through temporarily disregarding the digits in 'group1' and 'group2'. That's what gsub() is doing in the below code – it's replacing all digits in your 'group' types with "" , thus creating a conditional plot with three panels instead of four.

Now, the only thing left to do is define different colors for all levels(df$type) . The names() operation is not mandatory; it just helps to maintain an overview of the different factor levels and, in particular, their order.

# colors
clr = c("black", "black", "orange", "black")
names(clr) = levels(df$type) # for clarification only

# grouped scatterplot
xyplot(value~time | gsub("group[[:digit:]]", "group", type), df, group = type
       , layout = c(1, length(levels(df$type)) - 1), col = clr, cex = 1.2, pch = 20)