如何使用scale_linetype_manual修改ggplot图例名称并保持线型正确？

Let's work through this step by step—you're close, just need to adjust how you map your labels and linetypes properly in scale_linetype_manual.

The Core Problem #

When you tried to rename legend items to "current" and "origED3" and set linetypes, you probably either mixed up the order of values/labels, or didn't explicitly map your original type categories to the new labels and linetypes. Using "22" works as a quick fix because ggplot accepts numeric linetype codes, but we can make this cleaner and more reliable.

Correct Solution with Example Code #

Assuming your dataframe is named df, here's the proper way to set custom legend labels and linetypes without errors:

library(ggplot2)

# Example dataframe matching your structure
df <- data.frame(
  Year = rep(2015:2025, 2),
  mpft = rep(c("Group1", "Group1"), each = 11),
  value = c(runif(11, 0, 10), runif(11, 5, 15)),
  type = rep(c("original_cat1", "original_cat2"), each = 11), # Your original type categories
  index = 1:22
)

# Plot with correct legend labels and linetypes
ggplot(df, aes(x = Year, y = value, linetype = type)) +
  geom_line(linewidth = 1) + # Adjust line width for better visibility
  scale_linetype_manual(
    # Optional: Change the legend title (remove if you don't need it)
    name = "Scenario",
    # Map original type categories to your desired legend labels
    labels = c(
      original_cat1 = "current",
      original_cat2 = "origED3"
    ),
    # Map original type categories to valid ggplot linetypes
    values = c(
      original_cat1 = "solid",
      original_cat2 = "dashed"
    )
  ) +
  labs(x = "Year", y = "Value") +
  theme_bw()

Key Details to Avoid Errors #

1. Use Named Vectors for Labels & Values
   This is the most reliable approach—by naming each entry in labels and values with your original type category names, you eliminate any risk of mismatching linetypes to labels if the order of type levels changes.

2. Valid Linetype Options
   ggplot accepts these intuitive string values for linetypes (no need for numeric codes unless you need custom patterns):

   • "solid" (default solid line)
   • "dashed" (short dashes)
   • "dotted" (dots)
   • "dotdash" (dot-dash pattern)
   • "longdash" (long dashes)
   • "twodash" (double short dashes)

3. If You Don't Want Named Vectors
   If you know the exact order of your type levels (check with levels(df$type)), you can use unnamed vectors—just make sure the order of labels and values matches the order of type levels:

   scale_linetype_manual(
     labels = c("current", "origED3"),
     values = c("solid", "dashed")
   )
   

Why Your Temporary Fix Worked #

Numeric codes like "22" are valid in ggplot (they define custom dash-gap patterns), but they're less readable than the string names. Using the named vector approach ensures your plot stays maintainable and avoids accidental mismatches down the line.

内容的提问来源于stack exchange，提问作者Manfredo