回归绘图

R 中的 A/B 测试

Lauryn Burleigh

Data Scientist

线性拟合

cor.test(~ Enjoy + Time, data = pizza, 
         method = "pearson")
library(ggplot2)
ggplot(pizza, aes(x = Enjoy, 
                  y = Time)) + 
  geom_point()

散点图：x 为 Enjoy，y 为吃披萨所用时间，呈正相关。

ggplot(pizza, aes(x = Enjoy, 
                  y = Time)) + 
  geom_point() +
  geom_smooth(method = "lm")

散点图：x 为 Enjoy，y 为吃披萨所用时间；带蓝色最佳拟合线和置信区间。

线性预测

linear <- lm(Time ~ Enjoy, 
                data = Pizza) 
Enjoy  <- 12
topredict <-  data.frame(Enjoy) 
prediction <- predict(linear, 
                     newdata = topredict) 

ggplot(pizza, aes(x = Enjoy, y = Time)) + 
  geom_point() +
  geom_smooth(method = "lm") +  
  geom_hline(yintercept = prediction) + 
  geom_vline(xintercept = Enjoy)

散点图：x 为 Enjoy，y 为吃披萨所用时间；蓝色最佳拟合线和置信区间；在 x=12 与 y=6.24 处有相交的黑色线。

标示分组

ggplot(pizza, aes(x = Enjoy, y = Time, 
                  color = Topping)) + 
  geom_point() +
  geom_smooth(method = "lm")

散点图：x 为 Enjoy，y 为吃披萨所用时间；Pepperoni 的点和拟合线为粉色，Cheese 为蓝色。

标示分组

ggplot(pizza, aes(x = Enjoy, 
                  y = Time)) + 
  geom_point(aes(colour = Topping)) +
  geom_smooth(method = "lm")

散点图：x 为 Enjoy，y 为吃披萨所用时间，正相关；蓝色拟合线；Cheese 点为蓝色，Pepperoni 点为粉色。

逻辑回归曲线

ggplot(pizza, aes(x = Enjoy, 
                  y = EatAgain)) + 
  geom_point(aes(colour = Topping)) +
  geom_smooth(method = "glm",  
              method.args = 
                 list(family = 
                      "binomial"))

散点图：x 为 Enjoy，y 为 EatAgain，点仅在 0 或 1；带有弯曲的逻辑回归曲线。

逻辑回归预测

Enjoy  <- 12

ggplot(pizza2, aes(x = Enjoy, 
                   y = EatAgain)) + 
  geom_point(aes(colour = Topping)) +
  geom_smooth(method = "glm",  
              method.args = 
              list(family = 
                   "binomial")) +
  geom_vline(xintercept = Enjoy)

散点图：x 为 Enjoy，y 为 EatAgain，点仅在 0 或 1；Pepperoni 为粉色，Cheese 为蓝色；含逻辑回归曲线；x=12 处有黑色竖线。

Passons à la pratique !

R 中的 A/B 测试