Unit-4
Data Visualization
Using the diverse functionalities provided by R, one can create visually appealing data visualizations with only a few lines of code. Data visualization is an efficient technique of gaining insights about data through a visual medium.
- With the help of visualization techniques, humans can easily gain insights about the hidden patterns in data which might otherwise be neglected.
- Using data visualization, one can work with large datasets to efficiently obtain key insights about it.
R Visualization Packages
Following are some of the essential visualization packages in R Programming:
The basics of the graphics
Key elements of a statistical graphic:
- Data
- Aesthetic Mappings
- Geometric Objects
- Statistical Transformations
- Scales
- Coordinates system
- Faceting
Now, let us discuss each of them.
Aesthetic Mappings
- It controls the relation between data variables and graphics variables.
- Also, it helps to map the temperature variable of a data set into the X variable in a scatter plot.
- It helps to map the species of a plant into the colour of dots in graphics.
Geometric Objects
It shows each observation by a point using the aesthetic mappings that map two variables in the data set into the x,y variables of the plot.
Statistical Transformations
- It allows us to calculate and also perform a statistical analysis of the data in the plot.
- Also, the statistical transformation uses the data and approximates it by a regression line x,y coordinates.
- It counts occurrences of certain values.
Scales
It maps the data values into values in the coordinate system of the graphics device.
Coordinate system
We use it to plot the data.
- Cartesian
- Plot
Faceting
It splits the data into subgroups and draws sub-graphs for each group.
1. Controlling the Line Properties of a Chart
Line properties are crucial aesthetic elements in visualizations, particularly for line charts, where they help differentiate categories, emphasize trends or denote time series. In ggplot2, control over line properties like color, size and pattern relies on the distinction between aesthetic mapping and static setting.
A. Aesthetic Mapping for Data-Driven Variation
When a line property needs to vary based on a variable in the dataset (e.g., showing a different line type for each "Region"), the property is mapped inside the aes() function. This automatically generates a legend.
Example: Linking a data variable to a visual property.
B. Manual Scaling for Explicit Control
To override R's default choices and assign specific line properties to the groups, manual scale functions are added as layers to the plot.
Example: Overriding the default visual assignments generated by ggplot2 for mapped aesthetics.
Example-
Suppose x=c(1,2,3,4,5)
y=c(1,4,9,16,25)
plot(x,
y, type = "o", col = "blue", lty = 2, lwd = 2, pch = 17,
cex = 1.5,
main = "Line Properties Example", xlab = "X-axis", ylab = "Y-axis")
Where:
- col:
Line color ("red", "blue", "#FF5733" for hex
colors).
- lty:
Line type (1 for solid, 2 for dashed, etc.).
- lwd:
Line width.
- pch:
Point type (16 for filled circles, 17 for triangles, etc.).
- cex:
Marker size.
Output:
2. Using Keyword Arguments for Direct Static Styling
In ggplot2, "keyword arguments" are used for direct styling or static control of visual properties. This means setting a property to a constant value that applies to all elements within that geometric layer, irrespective of the data.
Example: Setting a property value outside the aes() function to apply a consistent style across the entire layer. This is used when the property does not need to be tied to a specific data variable and thus, no legend is generated.
In R: (Keyword arguments set properties statically)
ggplot(data, aes(x = Time, y = Value)) +
geom_line(color = "red", linewidth = 1.2)
Example:
Suppose x=c(1,2,3,4,5)
y=c(1,4,9,16,25)
plot(x,
y, type = "o", col = "darkgreen", lty = 1, lwd = 3, pch =
19,
main = "Keyword Arguments
Example", xlab = "X-axis", ylab = "Y-axis")
Here, the keyword arguments (type, col, lty, etc.)
are directly passed to the plot() function.
3. Using the Setter Methods
The concept of explicit setter methods (e.g., ax.set_title(), ax.set_xlabel()) is characteristic of the Object-Oriented (OO) plotting approach used in systems like Python's matplotlib. R's ggplot2 adopts a declarative, additive approach known as the Grammar of Graphics, where configuration is achieved by adding specific function layers, not by calling methods on an axis object.
Example: In ggplot2, the functionality of setters is replaced by adding specific non-data layers using the + operator.
In R-
# p is the plot object
p <- ggplot(data, aes(x = Month, y = Sales)) + geom_line()
# 1. Setting Labels (Equivalent to set_title, set_xlabel)
p <- p + labs(title = "Monthly Sales Report",
x = "Time Period",
y = "Total Revenue")
# 2. Setting Theme/Style (Equivalent to bulk style setters)
p <- p + theme_minimal() + # Use a built-in theme
theme(plot.title = element_text(face = "bold", size = 16)) # Custom styling
Example-
par(col =
"purple", lty = 3, lwd = 2) #
Set default line properties
plot(x,
y, type = "o", main = "Setter Methods Example")
4. Using the setp() Command
The setp() command is a specific utility function found in Python's matplotlib for setting multiple properties on a list of graphical "artists" (like lines or axes) using a MATLAB-style string/value pair format.
Example: Since ggplot2 is built on a declarative, layering system, there is no direct R equivalent of the setp() command. can use a combination of par() and with() to modify
multiple properties at once.
Example in R
(Bulk Setting Axis Text using theme()):
To set the color and size of all axis labels and tick marks globally:
p <- p + theme(axis.text = element_text(color = "grey50", size = 10),
axis.line = element_line(linewidth = 0.5))
Example:
# Set
multiple properties using par()
par(col =
"orange", lwd = 3, pch = 18)
plot(x,
y, type = "o", main = "Multiple Properties with par()")
ggplot2 objects into a single composite figure, essential for complex analysis where several views of the data are needed simultaneously.Concept: Graph composition relies on specialized R packages that treat each individual ggplot2 object as a component to be positioned within a unified grid layout.
A. Modern R Implementation: patchwork and cowplot
Modern R data visualization favors extension packages like patchwork (known for its intuitive, simple syntax) or cowplot.
B. The cowplot::plot_grid() Functioncowplot package provides the plot_grid() function, which allows explicit control over the layout, labels, and relative sizing of subplots.Example (Using cowplot to Combine Figures):
library(cowplot)
# p_rev and p_cust are two separate ggplot2 objects
p_rev <- ggplot(data_A, aes(x = Date, y = Revenue)) + geom_line() + labs(title = "Monthly Revenue")
p_cust <- ggplot(data_B, aes(x = Channel, y = Count)) + geom_bar(stat = "identity") + labs(title = "Customer Acquisition")
# Combine the plots side-by-side (ncol=2), label them, and adjust their width ratio
final_figure <- plot_grid(p_rev, p_cust,
ncol = 2,
labels = c("A. Revenue Trend", "B. Acquisition Channels"),
rel_widths = c(1.5, 1)) # Give p_rev 1.5 times the width of p_cust
Example-
Multiple Lines in One Plot:
y1 <- c(1, 4, 9, 16, 25)
y2 <-
c(1, 8, 27, 64, 125)
# Plot the first line
plot(x,
y1, type = "o", col = "blue", ylim = c(0, 130),
main = "Multiple Lines in One
Plot", xlab = "X-axis", ylab = "Y-axis")
# Add the second line
lines(x,
y2, type = "o", col = "red")
legend("topleft",
legend = c("y1 = x^2", "y2 = x^3"), col =
c("blue", "red"), lty = 1, pch = 16)
Using par(mfrow) for Subplots:
The par() function can be used to create subplots
within a single figure.
# Create
a 1x2 layout
par(mfrow
= c(1, 2))
# First plot
plot(x,
y1, type = "o", col = "blue", main = "Square
Plot", xlab = "X", ylab = "Y")
# Second plot
plot(x,
y2, type = "o", col = "green", main = "Cube
Plot", xlab = "X", ylab = "Y")
# Reset layout
par(mfrow
= c(1, 1))
Using layout() for Complex
Subplots:
For more control over layout:
layout(matrix(c(1,
2, 3, 3), nrow = 2, byrow = TRUE))
# Top left plot
plot(x,
y1, type = "o", col = "red", main = "Top Left")
# Top right plot
plot(x,
y2, type = "o", col = "blue", main = "Top Right")
# Bottom plot
plot(x,
y1 + y2, type = "o", col = "purple", main =
"Bottom")
The most used charts in data visualization:
- Scatter Plot
- Histogram
- Bar & Stack Bar Chart
- Box Plot
- Area Chart
- Heat Map
- Correlogram
2. Histogram
A histogram is used to plot a continuous variable. Also, It helps to break the data into bins and shows the frequency distribution of these bins. Thus, one can always change the bin size and see the effect it has on visualization.
3. Bar Chart
One can use bar charts to plot a categorical variable.
4. Box Plot
Box plots are used to plot an aggregation of categorical and continuous variables. It is also used for visualizing the spread of the data and detect outliers. Moreover, it shows five statistically significant numbers:
- Minimum
- 25th percentile
- Median
- 75th percentile and
- Maximum.
5. Area Chart
One can use it to show the continuity across a variable or data set. It is almost same as a line chart. Also, we can use it for time series plots. We can use it alternatively to plot continuous variables and analyze the underlying trends.
6. Heat Map
One can use it for the intensity of colours. It is also used to display a relationship between two or three or many variables in a two-dimensional image. Thus, it allows us to explore two dimensions of the axis and the third dimension by an intensity of colour.
7. Correlogram
One can use it to test the level of correlation and also among the variable available in the dataset. Thus, the cells of the matrix can be shaded or coloured to show the co-relation value.
Comments
Post a Comment