r - 如何从 ggplot2 的热图中排除异常值？

Question

嗨，我正在尝试在 map 上绘制每平方的平均转化率。效果很好。有问题的是只有很少记录的正方形经常达到接近 0 或 1 的极值。这使得图难以阅读。有没有办法排除未达到特定记录数的方 block ？或者设置颜色范围，例如 0.3 - 0.7？

代码:

library(ggplot2)
library(ggmap)
manila_map <- get_map("Manila,Philippines", zoom=11)

map <- ggmap(manila_map)
map + stat_summary_2d(
   geom = "tile", 
   data = data,
   fun = "mean",
   binwidth = 0.02,
   aes(x = lon, y = lat, z = requested),
   alpha = 0.4
) +
scale_fill_gradient2(low = "red", mid = "yellow", high = "#007f00", midpoint=0.5)

Answer 1

首先，将数据集中的离群值更改为NA

data$requested <- ifelse(data$requested <= 0.7 & data$requested >= 0.3, 
                         data$requested, NA)

然后，在 scale_fill_gradient() 中添加 na.value 使 NA 值成为中性色

scale_fill_gradient2(low = "red", mid = "yellow", high = "#007f00", midpoint=0.5, 
                     na.value = "grey50")