使用ggeffects绘制三向交互图时主题修改无法作用于所有分面

当使用ggeffects绘制包含4个terms的交互图时，plot()返回的是多个独立ggplot对象的组合体，直接用+添加主题只会作用于最后一个子图。以下是两种可行的解决方法：

方法1：预先设置ggeffects全局主题 #

通过ggeffects::set_theme()定义默认主题，让所有子图生成时自动应用该主题：

library(ggeffects)
library(ggplot2)

dat_iris <- iris
dat_iris$Petal.Size <- ifelse(dat_iris$Petal.Length <= 3, "small", "big")
model_iris <- glm(Sepal.Length ~ Sepal.Width*Species + Petal.Width + Petal.Size, data = dat_iris)

# 定义自定义主题
custom_theme <- theme_minimal() +
  theme(panel.grid.major = element_line(color = "#E5E5E5", linetype = "dashed", linewidth = 0.3),
        panel.background = element_rect(fill = "white", color = NA),
        plot.title = element_blank(),
        legend.title = element_text(face = "bold", size = 9),
        legend.text = element_text(size = 9),
        legend.position = "right",
        axis.title.x = element_text(face = "bold", size = 9), 
        axis.title.y = element_text(face="bold", size = 9),
        strip.background = element_rect(fill = "#F0F0F0", color = NA),
        strip.text = element_text(face = "bold", size = 9))

# 将自定义主题设为ggeffects默认主题
set_theme(custom_theme)

# 生成图形并添加其他标注、刻度设置
gg_iris <- plot(ggpredict(model_iris, terms=c("Sepal.Width", "Species", "Petal.Width", "Petal.Size")),  
                colors = c("#190C3EFF", "#9C2964FF", "#F8870EFF")) +
  xlab("Sepal Width") +
  ylab("Sepal Length") +
  guides(color = guide_legend("Species"), fill = guide_legend("Species")) +
  scale_y_continuous(breaks = 4:8) +
  scale_x_continuous(breaks = 2:5)

gg_iris

方法2：遍历子图逐个应用主题 #

将组合图拆分为单个ggplot对象，逐个修改后重新拼接：

library(ggeffects)
library(ggplot2)
library(patchwork)

dat_iris <- iris
dat_iris$Petal.Size <- ifelse(dat_iris$Petal.Length <= 3, "small", "big")
model_iris <- glm(Sepal.Length ~ Sepal.Width*Species + Petal.Width + Petal.Size, data = dat_iris)

# 生成原始组合图
gg_iris <- plot(ggpredict(model_iris, terms=c("Sepal.Width", "Species", "Petal.Width", "Petal.Size")),  
                colors = c("#190C3EFF", "#9C2964FF", "#F8870EFF"))

# 遍历每个子图，应用主题和其他设置
modified_plots <- lapply(gg_iris, function(p) {
  p +
    xlab("Sepal Width") +
    ylab("Sepal Length") +
    guides(color = guide_legend("Species"), fill = guide_legend("Species")) +
    theme_minimal() +
    theme(panel.grid.major = element_line(color = "#E5E5E5", linetype = "dashed", linewidth = 0.3),
          panel.background = element_rect(fill = "white", color = NA),
          plot.title = element_blank(),
          legend.title = element_text(face = "bold", size = 9),
          legend.text = element_text(size = 9),
          legend.position = "right",
          axis.title.x = element_text(face = "bold", size = 9), 
          axis.title.y = element_text(face="bold", size = 9),
          strip.background = element_rect(fill = "#F0F0F0", color = NA),
          strip.text = element_text(face = "bold", size = 9)) +
    scale_y_continuous(breaks = 4:8) +
    scale_x_continuous(breaks = 2:5)
})

# 重新拼接子图
wrap_plots(modified_plots)

原理说明 #

当ggpredict指定4个terms参数时，plot()会生成对应数量的独立ggplot对象并打包为列表/特殊组合体。常规的ggplot2语法+只能对单个ggplot对象生效，因此直接添加主题只会作用于最后一个子图。上述两种方法分别通过全局预设主题、逐个修改子图的方式，确保所有分面都应用指定的主题样式。

内容的提问来源于stack exchange，提问作者becbot