R geom_smooth 自定义最佳拟合线公式

Question

Geom_smooth formula

I am currently trying to fit a line to my data.

However, no fitted line is shown on the plot with none of the default methods of geom_smooth.

Here is the code that generated the plot:

p1 <- ggplot(data[ which(data$subject_id == "P121" & !is.na(data$col1)), ], aes(x = date, y= col1)) + 
  geom_point(colour= met.brewer("Isfahan1", 1)) + 
  geom_smooth(method = "stats::loess", se= FALSE) +
  theme(axis.title.x=element_blank(), axis.text.x=element_blank(),
                      axis.ticks.x=element_blank(), axis.title.y=element_blank(), axis.text.y=element_blank(),
                      axis.ticks.y =element_blank(), legend.title=element_blank()) +
  scale_color_manual(values=met.brewer("Isfahan1", 13))

This is the plot that is created by the code above I am aware that we can customize the formula for the model to be fitted, however, I'm new to R. Could you guide me on how to decide on the formula?

Answer 1

Your problem is using the string 'stats::' in the method parameter of the geom_smooth function.

See the code below

library(tidyverse)
library(MetBrewer)

i1 = c(21, 25, 27, 30:33, 35, 37, 38:41, 44, 46:48, 53, 54)
df = tibble(
  date = 1:55,
  col1 = ifelse(date %in% i1, 200, 100)
)

df %>%  ggplot(aes(date, col1)) + 
  geom_point(colour= met.brewer("Isfahan1", 1)) + 
  geom_smooth(formula = y~x, method = "loess", se= FALSE) + 
  theme(
    axis.title.x=element_blank(), 
    axis.text.x=element_blank(),
    axis.ticks.x=element_blank(), 
    axis.title.y=element_blank(), 
    axis.text.y=element_blank(),
    axis.ticks.y=element_blank(), 
    legend.title=element_blank()) +
  scale_color_manual(values=met.brewer("Isfahan1", 13))

You can also use another method, eg linear

df %>%  ggplot(aes(date, col1)) + 
  geom_point(colour= met.brewer("Isfahan1", 1)) + 
  geom_smooth(formula = y~x, method = "lm", se= FALSE) + 
  theme(
    axis.title.x=element_blank(), 
    axis.text.x=element_blank(),
    axis.ticks.x=element_blank(), 
    axis.title.y=element_blank(), 
    axis.text.y=element_blank(),
    axis.ticks.y=element_blank(), 
    legend.title=element_blank()) +
  scale_color_manual(values=met.brewer("Isfahan1", 13))

However, pay attention to the data you give to the ggplot function. If your date variable is not of numeric type, however, plot will be created without smooth path.

df = tibble(
  date = 1:55 %>% as.character(),
  col1 = ifelse(date %in% i1, 200, 100)
)

df %>%  ggplot(aes(date, col1)) + 
  geom_point(colour= met.brewer("Isfahan1", 1)) + 
  geom_smooth(formula = y~x, method = "loess", se= FALSE) + 
  theme(
    axis.title.x=element_blank(), 
    axis.text.x=element_blank(),
    axis.ticks.x=element_blank(), 
    axis.title.y=element_blank(), 
    axis.text.y=element_blank(),
    axis.ticks.y=element_blank(), 
    legend.title=element_blank()) +
  scale_color_manual(values=met.brewer("Isfahan1", 13))

As you can see in this example, I changed the value of the variable date to type character ( date = 1:55 %>% as.character() ) which made geom_smooth to not draw anything.