如何使用 ggplot2 为散点图绘制多条趋势线？

Posted 2023-02-16

【中文标题】如何使用 ggplot2 为散点图绘制多条趋势线？

【英文标题】：How can I draw multiple trendlines for scatterplots with ggplot2?

【发布时间】：2021-08-10 22:03:08

【问题描述】：

我正在尝试绘制一个基本的差异实现图（对于那些学习过经济学的人来说，您可能很熟悉）。所以我有两个独立的时间序列，涵盖 2015-2019 年。我想为两个时间序列绘制一条趋势线。问题是我为每个时间序列绘制了两个趋势线。第一个涵盖 1/2015-7/2017，第二个应涵盖 8/2017-12/2019。我已经设法通过以下使用 geom_smooth 来绘制线条：

    ggplot()+
      geom_line(data=timeseries2,aes(x=month,y=price_mean),color="red", size=1)+
      geom_line(data=timeseries1,aes(x=month,y=price_mean), color="dodgerblue4", size=1)+
      geom_smooth(data=timeseries2,aes(x=month,y=price_mean,color=time),method=lm,se=FALSE)+
      geom_smooth(data=timeseries1,aes(x=month,y=price_mean,color=time),method=lm,se=FALSE)+
      theme_classic()+
      labs(x="Month",y="Price")+
      geom_vline(xintercept =data1$month[32],linetype="dashed")

这是上面代码的结果：

这里的第一个问题是我不需要之前/之后的图例，我想用 Timeseries1/Timeseries2 图例替换它。第二个问题是 geom_smooth 趋势线的颜色。我希望 timeseries1 散点图具有相同的颜色，但之前和之后的条件都有单独的趋势线。这同样适用于其他时间序列。现在颜色是相同的，取决于时间而不是使用的时间序列。

【问题讨论】：

您的问题解决了吗？

【参考方案1】：

我创造了三种情节（我认为第二种是你的意思）

第一个情节：

library(dplyr)
library(ggplot2)

# Timeseries simulation
timeseries1 <- data.frame(
  month = seq(as.Date("2015/1/1"), as.Date("2019/12/1"), by = "month"),
  price_mean = sample(100:300, 60, replace = FALSE)
)

timeseries2 <- data.frame(
  month = seq(as.Date("2015/1/1"), as.Date("2019/12/1"), by = "month"),
  price_mean = sample(300:600, 60, replace = FALSE)
)

data1 <- rbind(timeseries1, timeseries2)

# dividing timeseries into two parts
timeseries1_B <- timeseries1 %>%
  filter(month >= as.Date(data1$month[32]))

timeseries1_A <- timeseries1 %>%
  filter(month <= as.Date(data1$month[32]))

timeseries2_B <- timeseries2 %>%
  filter(month >= as.Date(data1$month[32]))

timeseries2_A <- timeseries2 %>%
  filter(month <= as.Date(data1$month[32]))

#First plot
ggplot(data = timeseries1, aes(x = month, y = price_mean))+
  geom_line(data=timeseries1_A, color="dodgerblue4", size=1) + 
  geom_line(data = timeseries2_A, color = "dodgerblue4", size = 1) + 
  geom_line(data=timeseries1_B, color="red", size=1) + 
  geom_line(data = timeseries2_B, color = "red", size = 1) +
  geom_smooth(data=timeseries1_A,aes(x=month,y=price_mean, color = "dodgerblue4"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries2_A,aes(x=month,y=price_mean,color="dodgerblue4"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries1_B,aes(x=month,y=price_mean, color = "red"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries2_B,aes(x=month,y=price_mean,color="red"),method=lm,se=FALSE)+
  theme_classic()+
  labs(x="Month",y="Price")+
  geom_vline(xintercept =data1$month[32],linetype="dashed") +
  scale_color_manual(name = "Time", labels = c("Timeseries1", "timeseries2"), values = c("dodgerblue4","red"))

第二个情节：

#the same code as above then ...

#second plot:  
  
ggplot(data = timeseries1, aes(x = month, y = price_mean))+
  geom_line(data=timeseries1_A, color="dodgerblue4", size=1) + 
  geom_line(data = timeseries2_A, color = "red", size = 1) + 
  geom_line(data=timeseries1_B, color="dodgerblue4", size=1) + 
  geom_line(data = timeseries2_B, color = "red", size = 1) +
  geom_smooth(data=timeseries1_A,aes(x=month,y=price_mean, color = "dodgerblue4"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries2_A,aes(x=month,y=price_mean,color="red"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries1_B,aes(x=month,y=price_mean, color = "dodgerblue4"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries2_B,aes(x=month,y=price_mean,color="red"),method=lm,se=FALSE)+
  theme_classic()+
  labs(x="Month",y="Price")+
  geom_vline(xintercept =data1$month[32],linetype="dashed") +
  scale_color_manual(name = "Time", labels = c("Timeseries1", "Timeseries2"), values = c("dodgerblue4","red"))
    

第三个情节

# Timeseires simulation
timeseries1 <- data.frame(
  month = seq(as.Date("2015/1/1"), as.Date("2019/12/1"), by = "month"),
  price_mean = sample(100:300, 60, replace = FALSE)
)

timeseries2 <- data.frame(
  month = seq(as.Date("2015/1/1"), as.Date("2019/12/1"), by = "month"),
  price_mean = sample(300:600, 60, replace = FALSE)
)


data1 <- rbind(timeseries1, timeseries2)


ggplot()+
  geom_line(data=timeseries2,aes(x=month,y=price_mean),color="red", size=1)+
  geom_line(data=timeseries1,aes(x=month,y=price_mean), color="dodgerblue4", size=1)+
  geom_smooth(data=timeseries2,aes(x=month,y=price_mean,color="red"),method=lm,se=FALSE)+
  geom_smooth(data=timeseries1,aes(x=month,y=price_mean,color="dodgerblue4"),method=lm,se=FALSE)+
  theme_classic()+
  labs(x="Month",y="Price")+
  geom_vline(xintercept =data1$month[32],linetype="dashed") +
  scale_color_manual(name = "Time", labels = c("Timeseries1", "timeseries2"), values = c("dodgerblue4","red"))

它会创建如下内容：