I’ve run into this exact issue before—chart.Correlation is a convenient but heavily wrapped function, so standard base R plot parameters like pch, lty, or direct labels won’t work out of the box. Here’s why that happens and how to fix it:

Why Your Parameters Aren’t Working #

The chart.Correlation function uses custom panel functions (like panel.cor and panel.smooth) under the hood to draw the correlation matrix. These internal functions don’t inherit standard plot parameters directly—instead, the function exposes its own named arguments for controlling plot elements.

Solution 1: Use the Function’s Built-in Parameters #

First, check the official documentation for chart.Correlation—it has dedicated arguments for most styling needs:

• For scatter point symbols: Use point.pch instead of pch
• For point size: point.cex instead of cex
• For line styles in regression fits: You can pass styling arguments through the function’s ... parameter, but custom panels offer more control

Example code to change point symbols:

library(PerformanceAnalytics)
data(edhec) # Sample dataset

# Use point.pch to set scatter plot symbols
chart.Correlation(
  edhec[, 1:5],
  point.pch = 16, # Solid circles instead of default open circles
  point.cex = 1.2, # Slightly larger points
  col = "darkslategray" # Custom point color
)

Solution 2: Customize Panel Functions for Full Control #

If you need more flexibility (like changing regression line lty or customizing diagonal labels), you can override the default panel functions:

Customize Scatter Plots & Regression Lines #

# Define a custom panel function for scatter plots and fits
custom_smooth_panel <- function(x, y, col = plot$colors$points, 
                                pch = plot$point.pch, cex = plot$point.cex,
                                col.smooth = plot$colors$lines, lty = 2, ...) {
  # Draw points with your desired pch
  points(x, y, pch = pch, col = col, cex = cex)
  # Draw regression line with custom lty
  abline(lm(y ~ x), col = col.smooth, lty = lty)
}

# Call chart.Correlation with the custom panel
chart.Correlation(
  edhec[, 1:5],
  panel = custom_smooth_panel,
  point.pch = 17, # Triangle symbols
  lty = 2 # Dashed regression lines
)

Customize Diagonal Labels #

To modify the variable labels on the diagonal, override the text.panel parameter:

custom_text_panel <- function(x, y, labels, cex, ...) {
  text(x, y, labels, cex = 1.1, font = 2, col = "darkblue")
}

chart.Correlation(
  edhec[, 1:5],
  text.panel = custom_text_panel
)

Key Takeaway #

Stop trying to use base R plot parameters directly—chart.Correlation expects its own argument names or custom panel functions to modify plot elements. Always check the function’s documentation for supported arguments first, then fall back to custom panels for edge cases.

内容的提问来源于stack exchange，提问作者Kang