如何修改R语言investr包plotFit函数代码将散点改为柱状图？

Great question! The plotFit() function from the investr package doesn’t have a built-in parameter to swap scatter points for bar charts directly, but there’s a straightforward two-step workaround that preserves all your existing fit and confidence interval settings.

Step-by-Step Solution #

The core idea is to first draw your bar chart manually, then use plotFit() to overlay the fitted line and confidence interval on top of it. Here’s how to adapt your original code:

# Assume your NLS model is stored as 'nls_model' and your raw data is in a data frame 'df'
# with x variable 'x' and y variable 'y'

# 1. First draw the bar chart (adjust colors/limits to match your original plot)
barplot(height = df$y,        # Use your raw y-values as bar heights
        names.arg = df$x,     # Label bars with your original x-values
        col = "lightblue",    # Customize bar color as needed
        ylim = c(y1, y2),     # Keep your original y-axis limits
        xlim = c(x1, x2),     # Keep your original x-axis limits
        xaxp = c(0, 200, 10), # Keep your original x-axis tick settings
        ylab = "", 
        xlab = "", 
        main = "")

# 2. Overlay the fitted line and confidence interval with plotFit()
plotFit(nls_model, 
        interval = "confidence", 
        level = 0.95, 
        pch = NA,             # Disable scatter points (critical to avoid overlap!)
        shade = TRUE, 
        col.conf = "seagreen2", 
        col.fit = "green", 
        lwd.fit = 2, 
        n = 100, 
        add = TRUE)           # Tell plotFit to draw on top of the existing bar chart

Key Parameter Explanations #

• add = TRUE: This tells plotFit() to render the fit and interval on your existing bar chart instead of creating a new plot.
• pch = NA: This disables the default scatter point drawing, so you don’t get overlapping points on your bars.
• names.arg = df$x: Ensures your bar chart uses the original x-values from your data as labels, matching the axis scale from your original plotFit() call.

Note for Continuous X-Variables #

If your x-axis is continuous (not categorical), barplot() might shift bar positions slightly. For more precise alignment, use ggplot2 instead:

library(ggplot2)
library(investr)

# Get fitted values and confidence intervals
fit_data <- predictFit(nls_model, interval = "confidence", level = 0.95, n = 100)

# Plot with ggplot2
ggplot(df, aes(x = x, y = y)) +
  geom_col(fill = "lightblue") +
  geom_line(data = fit_data, aes(y = fit), color = "green", linewidth = 2) +
  geom_ribbon(data = fit_data, aes(ymin = lwr, ymax = upr), fill = "seagreen2", alpha = 0.3) +
  coord_cartesian(xlim = c(x1, x2), ylim = c(y1, y2)) +
  scale_x_continuous(breaks = seq(0, 200, 20)) + # Matches xaxp = c(0,200,10)
  labs(x = "", y = "", title = "")

内容的提问来源于stack exchange，提问作者willpkay