CODEWITHSUNDEEP
rggplot2predictionnlme
Mike
Mike

Reputation: 1161

Add fit lines from an nlme model to a ggplot, and include confidence intervals

I've used nlme to model my data like this:

library(nlme)
my_model <- lme(metric ~ group * time, random = "~1 | tech_rep", na.action = "na.omit", data = my_data)

I then want to plot my data, but include fit lines from the nlme model. I gather from some posts that this can be done with ggpredict (Extract prediction band from lme fit). I've managed to add the lines, but I'd also like the confidence interval to be displayed as a shared area around the line (as is default for geom_smooth). Setting se = TRUE doesn't seem to do anything.

ggpredict_data <- ggpredict(my_model, c("time", "group"), type = "re")
ggplot(ggpredict_data, aes(x = x, y = predicted, colour = group, fill = group)) +
    stat_smooth(method = "lm", se = TRUE, fullrange = TRUE) +
    geom_point(data = my_data,
               aes(x = time, y = metric, colour = group))

Can anyone help get the confidence interval working?

R, reading around, trying other posts

Upvotes: 0

Views: 148

Answers (1)

sometimes_sci
sometimes_sci

Reputation: 199

One way is to use different geoms to create the predicted line and confidence intervals separately:

ggplot(ggpredict_data, aes(x = x, y = predicted, colour = group, fill = group)) +
    geom_ribbon(aes(ymin = conf.low, ymax = conf.high), alpha = 0.5) + # for confidence intervals
    geom_line() + # for predicted line (aesthetics defined above)
    geom_point(data = my_data,
               aes(x = time, y = metric, colour = group))

Upvotes: 0

Related Questions