具有给定权重向量的ggplot平滑线glm模型

Question

我有如下数据：

numbers <- structure(list(density = c(1L, 4L, 10L, 22L, 55L, 121L, 210L, 
444L), females = c(1L, 3L, 7L, 18L, 22L, 41L, 52L, 79L), males = c(0L, 
1L, 3L, 4L, 33L, 80L, 158L, 365L), maleProp = c(0, 0.25, 0.3, 
0.181818181818182, 0.6, 0.661157024793388, 0.752380952380952, 
0.822072072072072), total = c(1L, 4L, 10L, 22L, 55L, 121L, 210L, 
444L)), .Names = c("density", "females", "males", "maleProp", 
"total"), row.names = c(NA, -8L), class = "data.frame")

我想用 glm 方法与 total 和 weight 有一条平滑的线。我试过了，

ggplot(numbers, aes(density, maleProp)) + geom_point() + 
  stat_smooth(method = "glm", 
              method.args = list(family = "binomial", 
                                 type = "response", 
                                 weights = "total"))

我收到错误，

Warning message:
Computation failed in `stat_smooth()`:
formal argument "weights" matched by multiple actual arguments

在这种情况下如何绘制平滑线？

Answer 1

如果您想使用带有您列出的参数的 glm，您可以创建一个包装函数，如下所示：

binomial_smooth <- function(...) {
  geom_smooth(method = "glm", method.args = list(family = "binomial"), ...)
}

您可以直接在 ggplot2 对象上使用：

ggplot(numbers, aes(density, maleProp)) + geom_point() + binomial_smooth(aes(weight = total))

Answer 2

作为 ggplot2 版本 2.2.1 的更新答案：

ggplot(numbers, aes(x = density, y = maleProp, weight = total)) +
    geom_point() + 
    stat_smooth(method = "glm", 
                method.args = list(family = "binomial"))

请注意，您可以使用 weight 作为 aes() 参数。

具有给定权重向量的ggplot平滑线glm模型

ggplot smooth line glm model with given vector of weights

r

ggplot2

glm