Implementation and Technical Analysis of Stacked Bar Plots in R

Data Preparation and Fundamental Concepts

To create stacked bar plots in R, proper data formatting is essential. Based on the Q&A data, users typically work with data frames containing multiple rows and columns, where columns represent categories (e.g., A to G) and values indicate metrics like duration in seconds. Data can be imported as follows:

dat <- read.table(text = "A   B   C   D   E   F    G
1 480 780 431 295 670 360  190
2 720 350 377 255 340 615  345
3 460 480 179 560  60 735 1260
4 220 240 876 789 820 100   75", header = TRUE)

This code uses the read.table function to create a data frame from a text string, with header = TRUE specifying the first row as column names. Data frames are standard structures in R for tabular data, forming the foundation for subsequent visualization tasks.

Creating Stacked Bar Plots with the Base Graphics System

R's base graphics system offers a straightforward method for generating stacked bar plots. The core function is barplot, which automatically handles stacking logic. Implementation code is simple:

barplot(as.matrix(dat))

The key step is converting the data frame to a matrix, as barplot expects matrix input by default. When a matrix is passed, each column is treated as a group, and each row as a stacked segment within that group. This approach is efficient for quick data exploration. In the resulting plot, the x-axis corresponds to categories A through G, the y-axis shows total duration, and different colors represent the original rows (i.e., data subsets).

Advanced Visualization with the ggplot2 Package

For more complex visualizations, the ggplot2 package provides a flexible and powerful solution. Unlike the base system, ggplot2 requires data in "long format," often necessitating data reshaping. This can be done using functions from packages like reshape2 or tidyverse. Here's an example using melt:

library(reshape2)
dat$row <- seq_len(nrow(dat))
dat2 <- melt(dat, id.vars = "row")

This code adds a row identifier column, then uses melt to transform data from wide to long format, where each observation occupies a row. The resulting data frame includes columns for row, variable (original column names), and value (numeric values).

Next, create the stacked bar plot with ggplot2:

library(ggplot2)
ggplot(dat2, aes(x = variable, y = value, fill = factor(row))) + 
  geom_bar(stat = "identity") + 
  labs(x = "\nType", y = "Time\n") + 
  theme_bw()

In the aes function, x is mapped to the category variable, y to the numeric values, and fill to the row identifier to produce stacking. geom_bar(stat = "identity") specifies that bar heights are based directly on data values, without statistical summarization. The labs function sets axis labels, and theme_bw applies a black-and-white theme. To hide the legend, add guides(fill = "none").

Technical Comparison and Best Practices

The base graphics method with barplot is suitable for quick, simple visualizations, with concise code and no need for data transformation. However, its customization options are limited, making tasks like adjusting colors, labels, or themes more cumbersome.

The ggplot2 approach, while requiring data reshaping, offers high flexibility and consistency. Through aesthetic mappings and a layered system, users can easily modify plot appearance, add annotations, or combine multiple plots. For example, to change the fill color scheme:

ggplot(dat2, aes(x = variable, y = value, fill = factor(row))) + 
  geom_bar(stat = "identity") + 
  scale_fill_brewer(palette = "Set3") + 
  theme_minimal()

Additionally, ggplot2 supports more complex data operations, such as preprocessing with dplyr or adding graphical elements like error bars.

In practice, the choice between methods depends on specific requirements. For exploratory data analysis or simple reports, the base system may suffice; for publication-quality graphics or complex dashboards, ggplot2 is often preferable. Regardless of the method, understanding data structure and visualization principles is crucial to avoid common pitfalls, such as incorrect data formats or mappings.

Common Issues and Solutions

When creating stacked bar plots, users might encounter issues. For instance, if data is non-numeric, barplot may fail to plot correctly; ensure conversion with as.numeric or check data import options. In ggplot2, if fill is mapped to a continuous variable instead of a factor, it might cause unexpected color gradients; using factor() to explicitly convert to a categorical variable can resolve this.

Another common problem is overlapping axis labels or crowded legends. In the base system, adjust with parameters like las and cex.axis; in ggplot2, fine-tune using theme functions and guide_legend.

In summary, stacked bar plots are powerful tools for displaying compositional data across categories. By mastering different implementation methods in R, users can select the most appropriate technique for their projects, creating clear and effective visualizations.