You need to enable JavaScript to run this app.
优惠活动
大模型
产品
解决方案
定价
更多

ggplot中geom_tile绘制混淆矩阵时顶部X轴标签与色块间间隙问题

ggplot中geom_tile绘制混淆矩阵时顶部X轴标签与色块间间隙问题

嘿,我帮你找到了几个导致顶部X轴标签和色块之间出现间隙的原因,咱们一步步来解决:

问题根源分析

  1. 错误的limits参数设置:你在scale_x_discrete里用了limits=truthClass,但truthClass是包含所有样本类别的重复向量,而limits需要的是唯一的类别水平列表,这个错误会让坐标轴范围计算出错,导致标签和色块错位。
  2. geom_tile的尺寸设置不合理:离散坐标轴上每个类别的间隔默认是1,你把height和width设为3,会让色块超出坐标轴的有效范围,进一步拉大标签和色块的距离。
  3. 轴标题的margin设置冗余:给axis.title.x加的上下左右20的margin,虽然不是核心问题,但会让整体布局的紧凑度变差。

修改后的完整代码

library(ggplot2)
library(dplyr)
library(stringr)

pred<- c(1,1,1,2,2,4,3,4,4,3,3,3)
truth<- c(1,1,1,2,2,2,3,3,3,4,4,4)

predClass<- pred
predClass[pred==1] = "Hi my friend"
predClass[pred==2] = "Hello my friend"
predClass[pred==3] = "Hey my friend"
predClass[pred==4] = "Hola my friend"

truthClass<- truth
truthClass[truth==1] = "Hi my friend"
truthClass[truth==2] = "Hello my friend"
truthClass[truth==3] = "Hey my friend"
truthClass[truth==4] = "Hola my friend"

truthClass <- as.factor(truthClass)
predClass <- as.factor(predClass)

tab <- table(truthClass, predClass)
tab <- tab / rowSums(tab)
tab <- as.data.frame(tab, stringsAsFactors = TRUE)

tab$predClass <- factor(tab$predClass, rev(levels(tab$predClass)))

ggplot(tab, aes(predClass, truthClass, fill = Freq)) +
  # 调整尺寸为贴合离散轴的0.9,也可以直接删除height和width用默认值
  geom_tile(colour = "#EBECEB", height=0.9, width=0.9) +
  geom_text(aes(label = ifelse(Freq == 0, "", scales::percent(Freq))),vjust = .5,
            alpha = 1, size = 7) +
  scale_fill_gradient(low = "white", high = "red") +
  # 替换limits为唯一的类别水平列表
  scale_x_discrete(labels = function(x)
    stringr::str_wrap(x, width = 9), limits=levels(truthClass),
    position = "top")+
  scale_y_discrete(labels = function(x)
    stringr::str_wrap(x, width = 9),limits=rev(levels(truthClass)))+
  labs(x = "True Class", y = "Prediction",
       fill = "Frequency") +
  theme_classic()+
  theme(legend.title = element element_text(colour = "black",
                                    size = 20, margin = margin(0, 20, 10, 0)),
        legend.text = element_text(size = 17),
        # 优化X轴标题的margin,只保留必要的底部间距
        axis.title.x = element_text(colour = "black", size = 20,
                                    margin = margin(10, 0, 0, 0)),
        axis.title.y = element_text(colour = "black", size = 20,
                                    margin = margin(0, 20, 0, 10)),
        axis.text.x = element_text(colour = "black",angle = 30, vjust = 0.6,
                                   size = 15),
        axis.text.y = element_text(colour = "black", size = 15),
        axis.line = element_line(colour = "#EBECEB",
                                 size = 1, linetype = "solid"),
        axis.ticks.length=unit(.25, "cm"))

关键修改点说明

  • 把scale_x_discrete的limits从truthClass改为levels(truthClass),确保坐标轴范围准确匹配类别水平;
  • 调整geom_tile的height和width为0.9(或直接删除这两个参数用默认值),让色块刚好贴合离散坐标轴的刻度区间;
  • 优化axis.title.x的margin设置,减少不必要的间距,让布局更紧凑。

这样调整后,顶部的X轴标签就会和色块紧密贴合,不会再有多余的间隙啦!

备注:内容来源于stack exchange,提问作者STEMQs

相关产品推荐
方舟 Agent Plan

超全模态模型 × Harness 升级,最新支持 Deepseek-V4.1-Flash、GLM-5.3 系列、Doubao-Seedream-5.0-pro、Kimi-K3 (部分), 限时 9.9 元起

最近更新时间:2026.04.23 10:54:03