ggplot2批量绘图时自动在图左上角放置线性回归结果的技术需求

Got it, let's fix this for you! The problem with your current approach is that it locks the regression statistics into the plot title, which means you can't set a separate, meaningful title for your visualization. Instead, we can use ggplot's annotate() function to place the adjusted R² and p-value directly in the top-left corner of the plot, leaving the title field free for your custom text.

Here's the updated, reusable function:

ggplotRegression <- function(fit) {
  require(ggplot2)
  
  # Pull the x and y variable names from the fitted model
  x_var <- names(fit$model)[2]
  y_var <- names(fit$model)[1]
  
  # Calculate and format the regression stats
  adj_r2 <- signif(summary(fit)$adj.r.squared, 2)
  p_val <- signif(summary(fit)$coef[2, 4], 2)
  stats_label <- paste0("Adj R² = ", adj_r2, "\nP = ", p_val)
  
  # Build the base plot with stats annotation
  ggplot(fit$model, aes_string(x = x_var, y = y_var)) +
    geom_point() +
    stat_smooth(method = "lm", color = "red") +
    # Place stats in top-left corner, aligned to the edge
    annotate("text",
             x = min(fit$model[[x_var]]),
             y = max(fit$model[[y_var]]),
             label = stats_label,
             hjust = 0, vjust = 1,
             size = 4, color = "black") +
    theme_bw() # Base theme included, but you can override later
}

Key Improvements: #

• Stats in Top-Left: Uses annotate() to position the R² and p-value at the top-left of the plot, based on the actual data ranges (so it won't drift off-screen no matter your variables).
• Custom Title Support: The title field is now free—you can add a meaningful, variable-specific title whenever you want.
• Flexible Formatting: Adjust the signif() argument (currently set to 2) if you want more/less decimal places for your stats.
• Reusable: Works with any linear model fit from lm()—just pass the model object, and it handles the rest.

Example Usage with mtcars: #

Let's test it with your 4-cylinder mtcars subset, adding a custom title and your preferred point styling:

# Load dplyr for filtering (if not already loaded)
require(dplyr)

# Filter 4-cylinder cars
disp_vs_wt_cyl4 <- mtcars %>% filter(cyl == 4)

# Fit the linear model
fit <- lm(disp ~ wt, data = disp_vs_wt_cyl4)

# Generate the plot with stats, custom title, and point style
ggplotRegression(fit) +
  geom_point(size = 3.74, colour = "#0c4c8a") +
  labs(title = "Displacement vs. Vehicle Weight (4-Cylinder Cars)")

This will give you a clean plot with the regression line, your styled points, a custom title, and the regression stats neatly tucked in the top-left corner—perfect for comparing any pair of variables in your dataset.

内容的提问来源于stack exchange，提问作者Tiptop