I apologize if this is a very basic question, but I’m having trouble defining colors when using plot_model in sjplot. As an example, I have run a simple multilevel regression model where 76 participants represent clusters. Each participant responded to a bunch of questions 16 times over the course of 4 days. The model goes
m<-lmer(happy ~ soc_smc + (soc_smc | id), data=data)
I want to plot the random slopes with plot_model:
p<-plot_model(m, type="pred", terms=c("soc_smc", "id"), pred.type="re", ci.lvl = NA)
Which works nicely, however sjplot base color palettes apparently have max 9 colors and I get 9 colored slopes and 67 grey ones.
I tried to create and use my own palette, e.g.
c1<-colors(distinct=T)
mycols<-sample(c1, 76)
p<-plot_model(m, type="pred", terms=c("soc_smc", "id"), pred.type="re", ci.lvl = NA, colors = "mycols")
However, I get the error "Error: Unknown colour name: mycols"
Same with colorspace
library(colorspace)
mycols<-qualitative_hcl(76)
p<-plot_model(m, type="pred", terms=c("soc_smc", "id"), pred.type="re", ci.lvl = NA, colors = "mycols")
Again plot_model says Error: "Unknown colour name: mycols"
I’m very inexperienced in creating color palettes so maybe I’m doing something wrong there, but why plot_model doesn’t seem to recognize the palette object? Or am I reading the error wrong?
I don’t actually need 76 distinctive colors (and that would be impossible too), however I’d like to get some color for every slope, and most importantly to know what I’m doing wrong.
>Solution :
You have to pass you color palette to the colors argument without quotes.
Using one of the default examples from sjPlot::plot_model:
library(sjPlot)
library(lme4)
#> Loading required package: Matrix
library(colorspace)
mycols <- qualitative_hcl(76)
m <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy)
plot_model(m,
type = "pred", terms = c("Days", "Subject"),
pred.type = "re", ci.lvl = NA, colors = mycols
)
