modified polar plot using ggplots or other alternative packages using R

Question

I am trying to create nice (!) polar plot with the following data.

gr1 <- c(0, 5, 15, 20, 30, 40)
gr3 <- c(0, 5, 10, 25, 40, 60, 80)
gr2 <- c(0, 15, 25, 30, 40)

df2<- data.frame (pos = c(gr1, gr2, gr3), group = c(rep(1, length(gr1)),
 rep(2, length(gr1)), rep(2, length(gr1))))



inner circle segment to mark
                   first tier, group 1, between 15, 20
                               group 3, between 5, 10
                                        between  40 to 60 
                   second tier, group 1, between 15, 20
                               group 3, between 5, 10
                                        between 10, 25
                                        between  40 to 60 

The angle between two lines between the interval between two pos.

there are more than two tier in my real data to plot.

The markers might be either segments or other marks such as filled colors.

Different group can color coded if possible.

Alternative with circle segment markings (preferred)

I tried it with ggplot2.

cx <- ggplot(df2, aes(x = pos, group = group) +   geom_bar(width = 1, colour = "black"))
Error in aes(x = pos, group = group) + geom_bar(width = 1, colour = "black") :
  non-numeric argument to binary operator
cx + coord_polar()

More trials:

df2$cpos <- cumsum (df2$pos)
cx <- ggplot(df2, aes(x = cpos, group = group) +   geom_bar(width = 1, colour = "black"))

Edits 1:

I think I might be terrible to explain the issue. Here is figure with explanation.

The pos (position) is a continuous scale, angle in polar should depend upon where value of position is. It will restart once one group is done. I did a trick to cumsum to this trick (I might be wrong).

Edits 2:

The following answer opened more thoughts into the question. This is improved version but color management is not working.

df2$cpos <- cumsum (df2$pos)
df2$y <- rep(3, length (df2$cpos))
cx <- ggplot(df2, aes(y = y, x = cpos))
cx + geom_bar(aes(stat = "identity",fill="yellow", colour = "yellow" )) + 
geom_point(aes(color = factor(group)), pch = 18, size = 2) + coord_polar() +
 scale_fill_manual(values=c("#CC6666", "#9999CC", "#66CC99"))+ ylim(0,3)

When I try to set ylim (2,3) so that they look like segments that do not work as well.

Error in data$y[data$y == -Inf] <- range$y.range[1] : 
  replacement has length zero

Answer 1

I am having a hard time understanding what you want to do (bar-plot counts frequencies, you probably know that). However here is a way to assign the groups into colors:

gr1 <- c(0, 5, 15, 20, 30, 40)
gr3 <- c(0, 5, 10, 25, 40, 60, 80)
gr2 <- c(0, 15, 25, 30, 40)

df2<- data.frame (pos = c(gr1, gr2, gr3), group = c(rep(1, length(gr1)),
 rep(2, length(gr2)), rep(3, length(gr3))))
df2$cpos <- cumsum (df2$pos)

cx <- ggplot(df2, aes(fill = factor(group), x = cpos))

cx + geom_bar(width = 1, colour = "black", position = "dodge")  + coord_polar()

If you want to get pos as frequencies, use the melt() function in reshape2.

If you want to use dots as in your example, could following approach work?

cx <- ggplot(df2, aes(y = group, x = cpos))

cx + geom_point(aes(color = factor(group))) + coord_polar() + ylim(0,3)

Anyway, you see the pattern? Make a plot with normal coordinates using x-axis as angle and y-axis as distance from the middle and just convert it to polar coordinates.

Answer to Edit2

I am still wondering whether you could make a plot that makes more sense, but maybe you have a reason to do this. Here is a plot that is closer to your example. It isn't perfect. Maybe the gurus can give you a better suggestion tomorrow once they arrive their offices. In a mean while you can look for more specifications from the links of this tread.

library(ggplot2)

gr1 <- c(0, 5, 15, 20, 30, 40)
gr3 <- c(0, 5, 10, 25, 40, 60, 80)
gr2 <- c(0, 15, 25, 30, 40)

df2<- data.frame (pos = c(gr1, gr2, gr3), group = c(rep(1, length(gr1)),
 rep(2, length(gr2)), rep(3, length(gr3))), y  = c(rep(1, length(gr1)),
 rep(2, length(gr1)), rep(2, length(gr1))))

df2$cpos <- cumsum (df2$pos)

cx <- ggplot(df2, aes(y = y, x = cpos))
cx + geom_point(aes(color = factor(group)), size = 4) + geom_line(aes(x = c(0,500), y = c(1)), color = "yellow") + 
geom_line(aes(x = c(0,500), y = c(2)), color = "blue") + scale_y_continuous(limits=c(0,2), breaks = c(0,1,2)) +
scale_x_continuous(labels = df2$pos, breaks = df2$cpos, limits = c(0,500)) + coord_polar() +
opts(panel.grid.major = theme_line(colour = "grey"), 
panel.grid.minor = theme_line(colour = "grey", linetype = "dashed"), 
panel.background = theme_rect(colour = "white"))