R语言基于tibble数据计算s16与v20差值制作发散条形图的实现方法
数据匹配与差值计算
当前数据集为上下拆分结构:上半段s16非空、对应v16有值,s20和v20缺失;下半段s20非空、对应v20有值,s16和v16缺失。按类别匹配成对记录、计算差值的操作流程:
- 拆分类别为两个独立子集:提取
s16非空的行,保留类别名与对应v16值;提取s20非空的行,统一类别命名规则,保留对应v20值 - 按类别名做连接,匹配同一类别下的两侧观测值
- 生成差值列:
diff_value = v16(s16侧) - v20(s20侧) - 过滤任意一侧数值缺失的无效行,保留同时具备两侧观测的有效类别
对应可直接运行的R代码:
library(dplyr) library(ggplot2) # 读入原始tibble raw_tibble <- tibble::tibble( s16 = factor(c("A","D","F","GG","GPS","FIND","LAB","N","SFT","P",NA,NA,NA,NA,NA,NA,NA,NA)), v16 = c(1017,152,15629,15437,1588,15853,3220,2165,5755,3455,NA,NA,NA,NA,NA,NA,NA,NA), s20 = factor(c(NA,NA,NA,NA,"GPS",NA,"LAB",NA,NA,NA,"U","F","GG","FIND","SFT","DP","BP",NA)), v20 = c(NA,NA,NA,NA,3650,NA,1548,NA,NA,NA,1058,13697,10893,16476,8464,3623,932,NA) ) # 拆分、连接、计算差值 s16_side <- raw_tibble %>% filter(!is.na(s16)) %>% select(category = s16, s16_val = v16) s20_side <- raw_tibble %>% filter(!is.na(s20)) %>% select(category = s20, s20_val = v20) paired_result <- left_join(s16_side, s20_side, by = "category") %>% mutate(diff_value = s16_val - s20_val) %>% filter(!is.na(diff_value))
运行后得到的有效匹配类别共6个,差值同时包含正、负取值,满足发散条形图的绘制要求。
发散条形图绘制
发散条形图的核心规则是所有条形以0值为中心轴,正向差值在轴一侧延伸,负向差值在另一侧延伸,通过颜色区分差值方向:
- 按差值大小升序排列类别,固定y轴顺序,避免条形位置无规则排布
- 以0值位置为基准绘制垂直参考线,强化发散视觉效果
- 用不同填充色区分正、负差值的条形
对应绘图代码:
# 固定类别顺序 paired_result <- paired_result %>% arrange(diff_value) %>% mutate(category = factor(category, levels = category)) # 绘制发散条形图 ggplot(paired_result, aes(x = diff_value, y = category, fill = diff_value > 0)) + geom_col(width = 0.6) + geom_vline(xintercept = 0, linetype = "solid", color = "gray30") + scale_fill_manual( values = c("TRUE" = "#2c7fb8", "FALSE" = "#d95f0e"), labels = c("TRUE" = "s16侧值更高", "FALSE" = "s20侧值更高") ) + labs( x = "s16侧值 - s20侧值的差值", y = "匹配类别", title = "两类观测值差值对比发散条形图" ) + theme_minimal() + theme(legend.position = "bottom")
如果需要在条形旁标注具体差值,可在绘图代码中追加文本层:
geom_text( aes(label = round(diff_value, 0)), hjust = ifelse(paired_result$diff_value > 0, 0.1, -0.1), size = 3 )
内容的提问来源于stack exchange,提问作者thewal
相关产品推荐
相关产品推荐

