CODEWITHSUNDEEP
rggplot2
Tyler Lane
Tyler Lane

Reputation: 89

Error with using side-by-side, nudged graphs in ggplot (facet_wrap and position_nudge)

I'm trying to create side-by-side plots with nudged data points (Odds ratios, with 95% CI error bars) in R using ggplot. Each time I try to combine them I get an error. Can anyone help me identify what I should do to change my code? This is the error I get:

Error in (~surv) + scale_x_continuous(breaks = seq(0, 4, 1)) : non-numeric argument to binary operator

To illustrate what I'm trying to do, see below a version I created using plot(), which you can see is fairly ugly: Example of plot I want to create using ggplot I've tried combining the facet_wrap and position_nudge based on the guidance in the J Stuart Carlton blog, but haven't been able to add a position_nudge. The error code above suggests that the problem is with facet_wrap section of my code. enter image description here

I've included code below describing how to replicate my dataset.

activity <- factor(rep(c("Good interaction", "Poor interaction",
                              "RTW plan"), times = 4))
surv <- factor(rep(c("T1", "T2"), each = 3, times = 2))
mod <- factor(rep(c("Crude", "Adjusted"), each = 6))
or <- c(1.72, 1.26, 2.39, 2.5, 1.34, 1.89, 1.14, 1.09, 2.02, 1.9, 1.1, 1.02)
low <- c(1.22, 0.74, 1.73, 1.74, 0.61, 1.35, 0.77, 0.61, 1.40, 1.22, 0.60, 0.68)
hi <- c(2.41, 2.16, 3.29, 3.6, 1.8, 2.64, 1.70, 1.94, 2.90, 2.95, 2.04, 1.54)
rtwc <- data.frame(activity, surv, mod, or, low, hi)

And here is the ggplot code I've been using:

ggplot(rtwc, aes(x = or, y = activity, colour = mod)) +
  geom_vline(aes(xintercept = 1), size = 0.25, linetype = "dashed") +
  geom_errorbarh(data = filter(rtwc, mod == "crude"), aes(xmax = hi, xmin = low), size = 0.5, height = 0.1, colour = "gray50", position = position_nudge(y = fix)) + 
  geom_point(data = filter(rtwc, mod == "crude"), aes(xmax = hi, xmin = low), size = 4, position_nudge(y = fix)) + 
  geom_errorbarh(data = filter(rtwc, mod == "Adjusted"), aes(xmax = hi, xmin = low), size = 0.5, height = 0.1, colour = "gray50", position = position_nudge(y = -fix)) + 
  geom_point(data = filter(rtwc, mod == "Adjusted"), size = 4, position = position_nudge(y = -fix)) +   
  geom_errorbarh(data = filter(rtwc, mod = "Adjusted")) + 
  facet_wrap = (~surv) +
  scale_x_continuous(breaks = seq(0, 4, 1)) + 
  coord_trans(x = "log10") +
  theme_bw() +
  theme(panel.grid.minor = element_blank())

Apologies if there is already a post on this question.

Upvotes: 1

Views: 997

Answers (1)

Z.Lin
Z.Lin

Reputation: 29125

Here are some code to get you started. I used position_dodge together with coord_flip to keep the error bar pairs away from each other:

ggplot(rtwc,
       aes(y = or, ymin = low, ymax = hi, x = activity, group = mod)) +
  geom_hline(yintercept = 1, size = 0.25, linetype = "dashed", colour = "grey") +
  geom_errorbar(width = 0.2, position = position_dodge(0.5)) +
  geom_point(aes(col = mod), position = position_dodge(0.5),
             size = 3) +
  scale_x_discrete(name = "") +
  scale_y_log10(name = "", breaks = seq(0, 4)) +
  scale_color_manual(name = "", values = c("red", "blue")) + # change colours here
  expand_limits(y = c(0.1, 4)) + #adjust x axis range here
  facet_wrap(~surv) +
  coord_flip() + 
  theme_bw() + #change look & feel here
  theme(panel.grid.minor = element_blank())

coefficient plot

For tweaks to the plot's look & feel, you can check out the available themes in ggplot here, & more themes in ggthemes here. Just please don't use the Excel 2013 theme. As the creator noted, its presence is for ironic purposes only.

For tweaks to the point colours, here's a handy reference for colours by name. Or you can use one of the palettes from RColorBrewer, viewable via RColorBrewer::display.brewer.all().

Upvotes: 2

Related Questions