[英]Merging legends in plotly subplot
我有幾個組,其中每個組都有幾個類,我測量了連續值:
set.seed(1)
df <- data.frame(value = c(rnorm(100,1,1), rnorm(100,2,1), rnorm(100,3,1),
rnorm(100,3,1), rnorm(100,1,1), rnorm(100,2,1),
rnorm(100,2,1), rnorm(100,3,1), rnorm(100,1,1)),
class = c(rep("c1",100), rep("c2",100), rep("c3",100),
rep("c2",100), rep("c4",100), rep("c1",100),
rep("c4",100), rep("c3",100), rep("c2",100)),
group = c(rep("g1",300), rep("g2",300), rep("g3",300)))
df$class <- factor(df$class, levels =c("c1","c2","c3","c4"))
df$group <- factor(df$group, levels =c("g1","g2","g3"))
並非數據中的每個組都具有相同的類,或者說不同的是每個組都有所有類的子集。
我正在嘗試為每個組生成R
plotly
密度曲線,按類進行顏色編碼,然后使用plotly
函數將它們全部組合成一個subplot
。
這就是我正在做的:
library(dplyr)
library(ggplot2)
library(plotly)
set.seed(1)
df <- data.frame(value = c(rnorm(100,1,1), rnorm(100,2,1), rnorm(100,3,1),
rnorm(100,3,1), rnorm(100,1,1), rnorm(100,2,1),
rnorm(100,2,1), rnorm(100,3,1), rnorm(100,1,1)),
class = c(rep("c1",100), rep("c2",100), rep("c3",100),
rep("c2",100), rep("c4",100), rep("c1",100),
rep("c4",100), rep("c3",100), rep("c2",100)),
group = c(rep("g1",300), rep("g2",300), rep("g3",300)))
df$class <- factor(df$class, levels =c("c1","c2","c3","c4"))
df$group <- factor(df$group, levels =c("g1","g2","g3"))
plot.list <- lapply(c("g1","g2","g3"), function(g){
density.df <- do.call(rbind,lapply(unique(dplyr::filter(df, group == g)$class),function(l)
ggplot_build(ggplot(dplyr::filter(df, group == g & class == l),aes(x=value))+geom_density(adjust=1,colour="#A9A9A9"))$data[[1]] %>%
dplyr::select(x,y) %>% dplyr::mutate(class = l)))
plot_ly(x = density.df$x, y = density.df$y, type = 'scatter', mode = 'lines',color = density.df$class) %>%
layout(title=g,xaxis = list(zeroline = F), yaxis = list(zeroline = F))
})
subplot(plot.list,nrows=length(plot.list),shareX=T)
我想解決的問題是:
使用plot_ly()
有點棘手,至少如果您想堅持使用color
參數從數據中生成多個跟蹤。
您需要考慮到您的類變量來定義legendgroup
組。 這legendgroup
但不會合並傳說物品放入一個(它只是它們分組)。
因此,為了避免圖例中的重復條目,您需要為那些要隱藏的圖例跟蹤設置showlegend = FALSE
。
由於您不使用循環(或 lapply)來創建子圖的單個軌跡,我們無法在生成圖時控制每個軌跡的可見性(通過上述color
參數 - 我們可以隱藏或顯示plot_ly
調用的所有跟蹤 - 通過add_trace
我們可以單獨控制每個跟蹤)。 因此,我僅為第一個圖設置showlegend = TRUE
並強制它通過虛擬數據顯示所有可用類。 請參閱以下內容:
set.seed(1)
df <- data.frame(value = c(rnorm(100,1,1), rnorm(100,2,1), rnorm(100,3,1),
rnorm(100,3,1), rnorm(100,1,1), rnorm(100,2,1),
rnorm(100,2,1), rnorm(100,3,1), rnorm(100,1,1)),
class = c(rep("c1",100), rep("c2",100), rep("c3",100),
rep("c2",100), rep("c4",100), rep("c1",100),
rep("c4",100), rep("c3",100), rep("c2",100)),
group = c(rep("g1",300), rep("g2",300), rep("g3",300)))
df$class <- factor(df$class, levels =c("c1","c2","c3","c4"))
df$group <- factor(df$group, levels =c("g1","g2","g3"))
library(dplyr)
library(ggplot2)
library(plotly)
plot.list <- lapply(c("g1","g2","g3"), function(g){
density.df <- do.call(rbind,lapply(unique(dplyr::filter(df, group == g)$class),function(l)
ggplot_build(ggplot(dplyr::filter(df, group == g & class == l),aes(x=value))+geom_density(adjust=1,colour="#A9A9A9"))$data[[1]] %>%
dplyr::select(x,y) %>% dplyr::mutate(class = l)))
p <- plot_ly(data = density.df, x = ~x, y = ~y, type = 'scatter', mode = 'lines', color = ~class, legendgroup = ~class, showlegend = FALSE) %>%
layout(xaxis = list(zeroline = F), yaxis = list(zeroline = FALSE)) %>%
add_annotations(
text = g,
x = 0.5,
y = 1.1,
yref = "paper",
xref = "paper",
xanchor = "middle",
yanchor = "top",
showarrow = FALSE,
font = list(size = 15)
)
if(g == "g1"){
dummy_df <- data.frame(class = unique(df$class))
dummy_df$x <- density.df$x[1]
dummy_df$y <- density.df$y[1]
p <- add_trace(p, data = dummy_df, x = ~x, y = ~y, color = ~class, type = "scatter", mode = "lines", showlegend = TRUE, legendgroup = ~class, hoverinfo = 'none')
}
p
})
subplot(plot.list, nrows = length(plot.list), shareX = TRUE)
另一種方法(避免虛擬數據解決方法)是在循環中(或通過 lapply)創建每個跟蹤,並根據項目的第一次出現控制它的圖例可見性。
此外,我認為應該可以使用?plotly::style
控制圖例項的?plotly::style
。 但是,我目前無法控制單個跟蹤。 我在這里提出了一個問題。
關於子圖的標題,請看這個。
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.