Comprehensive Guide to Plotting Multiple Columns in R Using ggplot2

Data Reshaping: From Wide to Long Format

In data visualization, we often need to plot multiple variables in the same graph. The ggplot2 package in R provides powerful plotting capabilities, but its standard usage requires data to be in long format. Original data is typically stored in wide format, where each variable occupies a separate column.

Consider the following example dataset:

A       B       C       D       Xax
0.451   0.333   0.034   0.173   0.22        
0.491   0.270   0.033   0.207   0.34    
0.389   0.249   0.084   0.271   0.54    
0.425   0.819   0.077   0.281   0.34
0.457   0.429   0.053   0.386   0.53    
0.436   0.524   0.049   0.249   0.12    
0.423   0.270   0.093   0.279   0.61    
0.463   0.315   0.019   0.204   0.23

Using the melt function from the reshape2 package effectively converts data from wide to long format:

library(ggplot2)
library(reshape2)

# Read data
d <- read.delim(textConnection(s), sep="")

# Reshape data to long format
d <- melt(d, id.vars="Xax")

The transformed data will contain three columns: Xax (identifier variable), variable (variable name), and value (variable value). This structure allows ggplot2 to easily handle visualization of multiple variables.

Plotting Multiple Variables in the Same Graph

Using the reshaped data, we can create a comprehensive graph containing all variables. ggplot2's aesthetic mapping system allows us to distinguish different variables by color:

ggplot(d, aes(Xax, value, col=variable)) + 
  geom_point() + 
  stat_smooth(method = "lm", se=FALSE)

In this code:

• aes(Xax, value, col=variable) defines x-axis as Xax, y-axis as value, and distinguishes variables by color
• geom_point() adds scatter plot layer
• stat_smooth(method = "lm", se=FALSE) adds linear regression lines for each variable without confidence intervals

Faceted Display: Separate Subplots

When dealing with numerous variables or requiring clearer comparisons, the faceting functionality can display each variable in separate subplots:

ggplot(d, aes(Xax, value)) + 
  geom_point() + 
  stat_smooth(method = "lm", se=FALSE) +
  facet_wrap(~variable)

The facet_wrap(~variable) function automatically creates multiple subplots based on the values in the variable column, with each subplot showing data points and regression lines for one variable.

Technical Deep Dive

The core advantage of data reshaping lies in its perfect alignment with ggplot2's grammar of graphics. ggplot2, based on the theory of graphical grammar, requires data to be in a specific structured format. Long-format data enables:

1. Unified Aesthetic Mapping: All variables share the same x and y aesthetic mappings
2. Simplified Layer Management: No need to repeat geometric objects for each variable
3. Enhanced Extensibility: Easy addition of new variables or modification of existing visualizations

Regression analysis visualization is achieved through stat_smooth(), which automatically calculates and plots regression lines for each group (defined by the color aesthetic). The method parameter specifies the regression method, "lm" indicates linear model, and the se parameter controls whether to display confidence intervals.

Common Issues and Solutions

Beginners often make the mistake of trying to add geometric objects separately for each variable, such as:

# Not recommended approach
ggplot(data=df) + 
  geom_point(aes(x=Xax,y=A)) + 
  geom_point(aes(x=Xax,y=B)) + 
  geom_point(aes(x=Xax,y=C)) + 
  geom_point(aes(x=Xax,y=D))

This approach not only results in verbose code but also makes maintenance and extension difficult. More importantly, statistical transformations like regression lines cannot be correctly applied to all variables.

The correct approach is always to first convert data to long format, then utilize ggplot2's grouping and faceting capabilities. This method ensures code conciseness, readability, and maintainability.

Extended Applications

Based on the same data reshaping principles, we can further extend visualizations:

# Add custom colors and themes
ggplot(d, aes(Xax, value, col=variable)) + 
  geom_point(size=2, alpha=0.7) + 
  stat_smooth(method = "lm", se=TRUE, alpha=0.2) +
  scale_color_manual(values=c("red", "blue", "green", "purple")) +
  theme_minimal() +
  labs(x="X-axis Variable", y="Y-axis Value", color="Variable Type")

This flexibility enables users to easily create professional-level scientific visualizations that meet various research and reporting requirements.