R语言Plotly绘图：仅让foo分类在图例中显示的实现方法

你之前的配置失效原因是：直接通过plot_ly按color = ~Category分组时，会自动为每一类生成独立trace，全局设置showlegend=F会作用于所有trace，需要单独对不需要显示图例的分类配置该参数，以下是两种可行实现方案：

方案1：拆分数据集分别添加trace（直观可控） #

library(dplyr)
library(plotly)
library(plyr)

df <- data.frame(
    Category = c('foo', 'bar', 'bar', 'foo'),
    x = c(2.1, 3.4, 4, 4),
    y = c(16, 21, 10, 17)
)

palette <- c("green", "red")

fig <- plot_ly() %>%
  # 添加foo分类trace，显示图例
  add_trace(
    data = filter(df, Category == "foo"),
    x = ~x, y = ~y, size = I(10), color = I("green"),
    symbol = I(15), hoverinfo = 'y', name = "foo", showlegend = TRUE
  ) %>%
  # 添加bar分类trace，隐藏图例
  add_trace(
    data = filter(df, Category == "bar"),
    x = ~x, y = ~y, size = I(10), color = I("red"),
    symbol = I(17), hoverinfo = 'y', name = "bar", showlegend = FALSE
  )

fig

方案2：基于原有代码修改，无需拆分数据集 #

通过style函数定向修改对应trace的showlegend属性即可，R默认Category因子按字母排序，bar对应第1条trace，foo对应第2条trace：

library(dplyr)
library(plotly)
library(plyr)

df <- data.frame(
    Category = c('foo', 'bar', 'bar', 'foo'),
    x = c(2.1, 3.4, 4, 4),
    y = c(16, 21, 10, 17)
)

palette <- c("green", "red")

fig <- plot_ly(
    data = df, x = ~x, y = ~y, size = 10, color = ~Category, colors = palette,
    symbol = ~Category, hoverinfo = 'y', symbols = c(15, 17)
) %>%
  # 隐藏bar对应trace的图例
  style(showlegend = FALSE, traces = 1)

fig

如果不确定分类对应的trace序号，可以运行plotly_json(fig)查看trace的name属性对应调整traces参数的取值。

内容的提问来源于stack exchange，提问作者Forklift17