Is it possible to combine the results of multiple univariate partial dependence plots on a single gr-CodePudding

I have a categorical random forest model with 3 classes of species assemblages and 18 variables, and I would like to build partial dependence plots for each class and each variable. I have the following lines of code for building four plots (how each class interacts with that variable "depth")

pdp1.1 <- partial(RFmodel, pred.var = "depth", plot=TRUE, which.class=1, train=train.df, plot.engine= "ggplot") 
pdp1.2 <- partial(model.rf.rf, pred.var = "a_bathym", plot=TRUE, which.class=2, train=train.df, plot.engine= "ggplot")
pdp1.3 <- partial(model.rf.rf, pred.var = "a_bathym", plot=TRUE, which.class=3, train=train.df, plot.engine= "ggplot")

Is there a way to make it so the results from each partial dependence plot can be combined into a single graph with the same x / y axis?

I'm happy to provide any additional information/code if it helps!

CodePudding user response：

Here is a solution for the basic ggplot fig objects, I don't if it will work for you. Meanwhile, I will try to make some PDP models and update this.

library(ggplot2)
#> Warning: package 'ggplot2' was built under R version 4.0.5
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(ggpubr)

p1 <- iris %>% ggplot(aes(Sepal.Length, Petal.Length))   geom_point()

p2 <- iris %>% ggplot(aes(Sepal.Length, Petal.Length))   geom_point()





fig <- ggarrange(p1   rremove("xlab")   rremove("ylab"), 
                 p2   rremove("xlab")   rremove("ylab"),  
                 common.legend = T)


fig %>% annotate_figure(left = text_grob("your y-axis", rot = 90),
                        bottom = text_grob("your x-axis"))

^{Created on 2022-02-19 by the reprex package (v2.0.1)}