了解为scale_fill_continuous_divergingx输入的参数以处理颜色边距

2024-04-28 • 问答

该问题是我上一个问题here的继续。

我有一个带有可用数据集的热图。数据集粘贴在下面：

library(ggplot2)
library(colorspace)

bigtest <- structure(list(x = c(-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,-8,8),y = c(0,5),z = c(1281.35043,576.76381,403.46607,363.28815,363.13356,335.04997,246.93314,191.56371,165.35087,136.33712,83.91203,107.5773,56.91087,56.91089,54.16559,54.18172,1841.60838,1098.66304,424.80686,363.52776,363.13355,335.04998,191.69473,83.91204,107.57729,56.91088,54.16421,54.16794,2012.52217,1154.7927,446.79023,363.31379,191.9613,83.91202,107.57731,54.1642,2077.10354,1217.43403,450.18301,363.44225,363.13357,363.13363,253.99753,218.43223,165.35014,107.57822,82.87399,2092.56391,1229.49925,451.15179,392.30728,363.13282,264.18944,218.4308,165.35044,83.92709,82.87353,82.87406,56.54491,2206.93318,1231.66411,457.37767,392.41558,363.13283,335.06272,191.95211,136.35211,112.12755,82.73634,82.87418,54.16421)),row.names = c(NA,-102L),class = c("tbl_df","tbl","data.frame"))

我正在使用以下代码段生成热图：

ggplot(bigtest,aes(x = x,y = y)) +
  geom_tile(aes(fill = z)) +
  scale_fill_continuous_divergingx(palette = 'RdBu',rev = TRUE,mid = 347.48,l3 = 54,p3 = 2206,p4 = 325)

了解为scale_fill_continuous_divergingx输入的参数以处理颜色边距

我从图形中期望的是，白色将以某个特定值为中心，而其他渐变将基于该值而高于或低于该值。但是，通过使用不同的参数，似乎我无法完全理解参数l3，p3和p4的含义。当我查看此功能的documentation时，它建议自定义比例的参数来自colorspace包中的divergingx_hcl函数。

在查看divergingx_hcl documentation时，指出它们与不同的输入参数相对应。我完全迷路了，完全不知道这是什么。任何有关帮助我围绕这些参数（不仅是l3，p3和p4，还包括其他参数）的指导都将不胜感激。

^{由reprex package（v0.3.0）于2019-11-07创建}

首先，将所有颜色都指定为HCL（色相，色度，亮度），其对应于颜色的类型（红色，绿色，蓝色等），颜色的鲜艳程度（低色度为灰色，高色度）颜色非常鲜艳），以及颜色的亮度（高亮度是白色，低亮度是黑色）。

参数l3表示色标一端的颜色亮度分量。（l1是另一端的亮度，l2是中间的亮度。）亮度从0到100。因此，如果希望末端的颜色更暗，请设置亮度降低。参数p3和p4是控制颜色从中点过渡到端点的速度的指数。通常，接近0的值表示过渡快，而大于1的值则过渡慢。您不太可能希望p3或p4的值大于10。

要获取调色板的默认参数，可以使用divergingx_palettes()命令：

library(colorspace)
divergingx_palettes('RdBu')
#> HCL palette
#> Name: RdBu
#> Type: Diverging (flexible)
#> Parameter ranges:
#>  h1 h2  h3 c1 c2 c3 l1 l2 l3  p1
#>  20 NA 230 60  0 50 20 98 15 1.4

^{由reprex package（v0.3.0）于2019-11-07创建}

这表明l3指定的终点颜色已经很暗。将l3从15更改为0会使它变暗一些，但幅度不大。此外，未指定p2，p3和p4，这意味着它们全部取自p1，因此为1.4。因此，颜色插值比线性插值要慢一些。

有了这些知识，以下示例应该是有意义的。要了解更多信息，我建议阅读colorspace网站上的各种文章：http://colorspace.r-forge.r-project.org/

首先数据：

library(ggplot2)
library(colorspace)

bigtest <- structure(list(x = c(-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,-8,8),y = c(0,5),z = c(1281.35043,576.76381,403.46607,363.28815,363.13356,335.04997,246.93314,191.56371,165.35087,136.33712,83.91203,107.5773,56.91087,56.91089,54.16559,54.18172,1841.60838,1098.66304,424.80686,363.52776,363.13355,335.04998,191.69473,83.91204,107.57729,56.91088,54.16421,54.16794,2012.52217,1154.7927,446.79023,363.31379,191.9613,83.91202,107.57731,54.1642,2077.10354,1217.43403,450.18301,363.44225,363.13357,363.13363,253.99753,218.43223,165.35014,107.57822,82.87399,2092.56391,1229.49925,451.15179,392.30728,363.13282,264.18944,218.4308,165.35044,83.92709,82.87353,82.87406,56.54491,2206.93318,1231.66411,457.37767,392.41558,363.13283,335.06272,191.95211,136.35211,112.12755,82.73634,82.87418,54.16421)),row.names = c(NA,-102L),class = c("tbl_df","tbl","data.frame"))

现在的情节：

ggplot(bigtest,aes(x = x,y = y)) +
  geom_tile(aes(fill = z)) +
  scale_fill_continuous_divergingx(
    palette = 'RdBu',rev = TRUE,mid = 347.48
  )

ggplot(bigtest,mid = 347.48,p3 = .2,p4 = .2
  )

ggplot(bigtest,l3 = 0,p4 = .2
  )

^{由reprex package（v0.3.0）于2019-11-07创建}

了解为scale_fill_continuous_divergingx输入的参数以处理颜色边距

tangfeng1212 回答：了解为scale_fill_continuous_divergingx输入的参数以处理颜色边距

大家都在问