绘制 R 中不同数据集的拟合概率

Question

我想在图表中绘制回归的拟合回归线，其中还绘制了 -1 和 1 范围内的分箱变量。

lost <- ggplot(data = data_2, 
  aes(x=log_rank_ratio, y=quit, color = as.factor(favorite))) +
  geom_point(size=2) +
  theme_bw() + 
  xlab("Log Rank Ratio") +
  ylab("Observed Likelihood of Quitting Mid-Match") +
  geom_vline(xintercept=0, linetype="dashed", alpha=.7)

给出了下图（散点图中的x变量被分箱（data_2）：

但是，我想要添加的是绘制以下回归的拟合线（data_1 中未分箱的 x 值）：

fit1 <- lm(quit ~ favorite + log_rank_ratio + log_rank_ratio^2 + log_rank_ratio^3 , data = data_1)

应该给出这样的东西：

Answer 1

如果您确实想拟合三次多项式，请使用

geom_smooth

与

method = lm

和

formula = y ~ poly(x, 3)

。您还需要将

data = data_1

传递给

geom_smooth

，如果您希望线条为黑色，请使用

color = "black"

outside

aes

和

group = factor(favorite)

inside

aes

:

library(ggplot2)

ggplot(data = data_2, 
       aes(x = log_rank_ratio, y = quit, color = as.factor(favorite))) +
  geom_point(size = 2) +
  geom_smooth(data = data_1, method = lm, formula = y ~ poly(x, 3), 
              se = FALSE, aes(group = as.factor(favorite)), color = "black") +
  theme_bw() + 
  xlab("Log Rank Ratio") +
  ylab("Observed Likelihood of Quitting Mid-Match") +
  geom_vline(xintercept = 0, linetype = "dashed", alpha = 0.7)

^{创建于 2023-08-05，使用 reprex v2.0.2}

---

使用的数据（从问题中大致推断）

set.seed(1)

data_2 <- data.frame(log_rank_ratio = c(seq(-2, -0.2, 0.2), seq(0.2, 2, 0.2)),
                     quit = c(189, 177, 191, 212, 209, 213, 225, 249.3, 249.2,
                              249.1, 249.7, 280, 275, 305, 281, 280, 305, 325,
                              326, 275)/1e4,
                     favorite = rep(0:1, each = 10))

data_1 <- data.frame(log_rank_ratio = rep(data_2$log_rank_ratio, each = 10),
                     quit = c(t(replicate(10, rnorm(20, data_2$quit, 0.005)))),
                     favorite = rep(0:1, each = 100))