Pixel-Oriented Visualization Techniques

Pixel-oriented visualization techniques display data by assigning one pixel to each

data point, making it possible to show a lot of information on the screen. These
techniques are helpful for analyzing very large datasets.

Classification of Techniques
1. Query-Independent Techniques
• What it means: These techniques show the data as it is, without focusing on
any specific question or query.
• Examples:
o Peano-Hilbert Curve: A continuous, squiggly line connecting data
points.
o Morton Curve (Z-Curve): A zigzag line connecting data points.
• Best for: Large datasets with a natural order, like time-series data.

2. Query-Dependent Techniques
• What it means: These techniques highlight data based on how closely it
matches a specific query or question.
• Examples:
o Snake-Spiral: Puts the most relevant data in the center of the screen.
o Snake-Axes: Adds direction to show if data is increasing or
decreasing.
• Best for: Exploring data interactively based on a query.

3. Grouping Techniques
• What it means: These group all variables (details) of one data point into a
single block or area.
 • Key Features:
o Uses separate windows or areas for each data point.
o Makes it easier to compare individual records.
• Best for: Small datasets or focused analysis.

Details of Each Technique

1. Screen-Filling Curve Techniques
• Purpose: Arrange data on a 2D screen to reveal patterns like clusters or
trends.
• Key Points:
1. Converts one-dimensional data into a two-dimensional visual form.
2. Peano-Hilbert Curve: Keeps nearby data points close together.
3. Morton Curve: Easier to follow due to its regular Z-shape.
4. Helps in finding clusters (groups of similar data).
5. Best for: Time-series data like stock prices.
6. Challenge: Fixed structure may not represent data meaningfully.
Peano-Hilbert Algorithm
The Peano-Hilbert algorithm creates a squiggly path (curve) that goes through
every tiny box (pixel) in a big square. It is helpful to keep nearby data points close
to each other in the display.
Steps (Beginner-Friendly):
1. Start with a Big Square:
o Imagine a big square divided into 4 smaller squares.
2. Make a Path:
o Draw a line connecting the 4 smaller squares in a single continuous
way.
o The path looks like an "S" or "U" shape.
 3. Divide Again:
o Take each smaller square and divide it into 4 even smaller squares.
4. Adjust the Path:
o Change the line inside the smaller squares to make sure it connects
smoothly.
5. Repeat Until Pixels:
o Keep dividing and adjusting the path until the square is so small it’s
just a single pixel.
6. Draw the Final Path:
o The result is a smooth path that visits every pixel in the square
without jumping around.
Key Ideas:
• Why Use It?
o It keeps nearby data points close in the picture, so patterns in data are
easier to spot.
• Challenge:
o It’s harder to follow the squiggly path visually, especially in large
grids.

Morton Algorithm (Z-Curve)

The Morton algorithm (or Z-curve) draws a simpler path in the shape of a "Z" to
connect tiny boxes (pixels) in a square. It is easier to understand than the Peano-
Hilbert algorithm.
Steps (Beginner-Friendly):
1. Start with a Big Square:
o Divide the big square into 4 smaller squares.
2. Draw a Z Shape:
o Connect the 4 smaller squares in the shape of a "Z."
 o Start at the top-left, move to the top-right, then go to the bottom-left,
and finally the bottom-right.
3. Divide Again:
o Take each smaller square and split it into 4 even smaller squares.
4. Repeat the Z-Shape:
o Follow the same "Z" path in each tiny square.
5. Stop at Pixels:
o Keep dividing until the squares become individual pixels.
6. Draw the Final Path:
o The result is a zigzag path that covers every pixel in the square.
Key Ideas:
• Why Use It?
o It’s simple to calculate and easy to follow visually.
• Drawback:
o It might not keep nearby data points as close as the Peano-Hilbert
algorithm does.

Comparison (Key Differences):

1. Peano-Hilbert Curve:
o Makes a smooth, continuous path.
o Keeps nearby data very close together.
o Harder to visually trace the path.
2. Morton (Z-Curve):
o Uses a straightforward "Z" pattern.
o Easier to compute and follow.
o May not group nearby data as tightly.
Would you like diagrams or examples to make this even clearer?

2. Recursive Pattern Technique

• Purpose: Arrange pixels in a way that reflects natural groupings in the data.
• Key Points:
1. Divides data into levels (e.g., days, weeks, months).
2. Customizable based on user preferences.
3. Maintains patterns from larger to smaller data levels.
4. Example: Grouping daily stock prices into monthly trends.
5. Makes complex data structures easier to understand.
6. Challenge: Setting up recursion levels requires attention.

3. Snake-Spiral Technique
• Purpose: Show the most relevant data in the center.
• Key Points:
1. Sorts data based on its relevance to a query.
2. Displays data in a spiral pattern, starting from the middle.
3. Aligns related data across different windows for comparison.
4. Example: Spotting clusters in customer or test data.
5. Improves over traditional spirals by allowing customization.
6. Challenge: Can be tricky to understand spiral arrangements.

4. Snake-Axes Technique
• Purpose: Add direction to the visualization for better understanding.
 • Key Points:
1. Splits data into 4 parts based on increasing or decreasing trends.
2. Assigns two variables to horizontal and vertical axes.
3. Helps identify patterns like clusters more clearly.
4. Example: Comparing sales vs. expenses in a business dataset.
5. Requires user input to choose meaningful variables for axes.
6. Challenge: Reduces data density due to partitioning.

5. Grouping Technique
• Purpose: Group all details of one data point together.
• Key Points:
1. Displays all variables of a data item in one small area.
2. Example: Showing multiple features of a customer in one place.
3. Reduces clutter by focusing on selected data points.
4. Can use spiral or recursive patterns inside groups.
5. Requires more pixels per data point, limiting large datasets.
6. Best for: Focused searches on small datasets.
Geometric Projection
Geometric Projection Visualization Techniques
What are they?
Geometric projection visualization techniques represent high-dimensional data
(data with many variables) by projecting it into a 2D or 3D space for easier
visualization and analysis.
Key Features:
1. Dimensional Reduction:
o Simplifies data by projecting multiple variables into fewer dimensions
(like 2D or 3D).
o Helps to make sense of complex, high-dimensional datasets.
2. Focus on Relationships:
o Highlights patterns, clusters, or trends between data points.
3. Techniques Include:
o Scatter Plots: For 2D or 3D representation.
o Radial Projection (Star Coordinates): Projects data outward from a
central point.
o Parallel Coordinates: Represents each variable as a vertical line and
data points as crossing lines.
4. Easy Comparisons:
o Geometric visualizations make it easier to compare data points or
detect anomalies.
5. Interactive Visualization:
o Often used with tools that allow zooming, rotating, or exploring
specific sections.

(see notes for classification)

Radial Projection (Star Coordinates)
What is it? Radial projection is a way to visualize data with many variables
(multivariate data) by turning each variable into a line (axis) that spreads outward
from the center, like the rays of a star.

Beginner-Friendly Explanation
 1. Think of a Wheel:
o Imagine a wheel or a circle with lines (like spokes) coming out from
the center.
o Each line (spoke) represents one variable in your data.
2. Place the Data Points:
o For each data point, check its value for every variable.
o Place it along the line for that variable, closer to the center for smaller
values and farther for larger values.
3. Connect the Dots:
o Combine the positions on all lines for a data point to create a shape.
o For example, one data point might form a triangle, while another
forms a pentagon.
4. Compare Patterns:
o Similar data points will create similar shapes.
o Differences in shapes reveal how the data points vary across the
variables.

Key Features
1. Axes Radiating Outward:
o The lines spread outward like rays from a single point.
2. Data Representation:
o Each point represents a data item based on its values for all variables.
3. Easy Comparison:
o Shapes and their sizes help compare different data points.

Use Case
• Best For:
 o Small to medium datasets with many variables.
o Example: Comparing the health metrics (weight, height, blood
pressure) of 10 people.
• Purpose:
o Detect patterns, like clusters or outliers, in the data.
Parallel Coordinate Visualization Technique
What is it?
Parallel coordinate visualization is a method to represent data with many variables
(multivariate data) by drawing it as lines across multiple vertical axes, arranged
side by side.

Beginner-Friendly Explanation
1. Think of Multiple Lines:
o Imagine several vertical lines placed next to each other.
o Each vertical line (axis) represents a variable in your data.
2. Mark the Data:
o For a single data point, find its value for each variable.
o Place a mark (point) on each vertical axis at the correct value.
3. Connect the Dots:
o Connect these marks with a straight line that crosses all the axes.
o This line represents the data point across all variables.
4. Repeat for All Data:
o Repeat the process for every data point in your dataset.
o The result is a web of lines, each representing one data point.

Key Features
1. Side-by-Side Axes:
 o Vertical axes for variables are lined up in a row.
2. Data as Lines:
o Each data point becomes a line crossing all the axes.
3. Identify Patterns:
o Lines that are close or follow a similar path represent similar data
points.
4. Find Relationships:
o Crossing lines can reveal trends, clusters, or outliers in the data.

Use Case
• Best For:
o Medium-sized datasets with multiple variables.
o Example: Comparing cars based on mileage, horsepower, weight, and
price.
• Purpose:
o Spot trends, clusters, or unusual patterns in data.

Simple Example
If you're analyzing students' grades in 4 subjects:
• Each subject is a vertical axis.
• A student's marks are plotted on these axes and connected as a line.
• By comparing the lines, you can see who performed similarly or differently
across subjects.
Bar Graph
• What is it?
A bar graph represents data using rectangular bars where the length of the
bar corresponds to the value.
 • Key Features:
1. Bars can be vertical or horizontal.
2. Used to compare values between different categories.
3. Easy to read and interpret.
• Use Case:
Comparing sales of products, population in cities, or test scores of students.
• Example:
Showing the number of books read by students in different grades.

Pie Chart
• What is it?
A circular chart divided into slices, where each slice represents a proportion
of the whole.
• Key Features:
1. Useful for showing percentages or parts of a whole.
2. The size of each slice depends on its percentage.
• Drawback:
1. Difficult to compare similar-sized slices.
2. Not effective for datasets with many categories.
• Use Case:
Visualizing the market share of companies or the budget distribution in a
project.

Donut Chart
• What is it?
A variation of a pie chart with a hole in the middle, often used for aesthetic
purposes or to display total value in the center.
• Key Features:
 1. Similar to pie charts but less cluttered.
2. Better for visualizing differences in proportions.
• Drawback:
Faces the same issues as pie charts (difficult comparison of slices).
• Use Case:
Showing percentage distribution while emphasizing the total, like total
expenses broken down into categories.

Line Graph
• What is it?
A graph that uses lines to connect data points, showing trends over time or
continuous data.
• Key Features:
1. Ideal for showing changes over time.
2. Easy to spot trends, increases, or decreases.
3. Can handle multiple data series on the same graph.
• Use Case:
Plotting a company’s revenue growth over months or tracking temperature
changes during a day.
• Example:
A line showing the number of visitors to a website over a week.

Mosaic Plot
• What is it?
A rectangular visualization divided into tiles, where each tile represents a
group based on multiple variables.
• Key Features:
1. Each rectangle’s size represents the frequency or proportion of the
data.
 2. Used for categorical data (e.g., gender vs. education level).
3. Helps explore relationships between multiple categories.
• Use Case:
Analyzing the relationship between age, income, and spending habits.

• Example:
A mosaic plot showing voting patterns based on age groups and income
levels.
Icon Based Visualization Techniques
Icon-Based Visualization is a type of data visualization technique that uses simple
pictures or icons to represent data values. This makes it easier for people to
understand complex data by associating each piece of information with a
recognizable image or symbol.
Here, we'll explore two specific types of icon-based visualizations: Chernoff
Faces and Stick Figures.

1. Chernoff Faces:
Chernoff Faces are a way to represent data using human-like faces. Each face
represents a set of data values, where different features of the face (like the shape
of eyes, nose, mouth, and face) are used to show different values.
Key Points about Chernoff Faces:
• Data Representation: The data is represented by facial features like:
 o Eyes: Represent one feature of the data.
o Nose: Represents another feature.
o Mouth: Represents yet another data point.
o Other features like the face shape, size, and position of elements are
also adjusted according to data.
• How it works:
o Each facial feature corresponds to a specific piece of data (e.g., the
width of the eyes could represent a certain numerical value).
o By adjusting these features, you can visually compare how one set of
data differs from another.
• Advantages:
o It's easy to understand because people are familiar with human faces.
o It can show multiple data points at once in a compact form.
• Example: Suppose you're comparing the performance of three products:
o Product A: Eyes are wide, mouth is smiling.
o Product B: Eyes are smaller, mouth is neutral.
o Product C: Eyes are very small, mouth is frowning. Each face would
tell you about each product's features like price, quality, and user
rating based on how the facial features change.
How it helps:
• Quick Interpretation: You don't need to look at complex numbers or
graphs. A quick glance at the faces helps you understand how things
compare.
• Appeals to Visual Learners: This method helps those who find it hard to
interpret numeric data but are comfortable interpreting visual patterns.
2. Stick Figures:
Stick Figures use simple human-like drawings made of sticks to represent data.
Like Chernoff Faces, each stick figure will represent data points, and different
parts of the figure (such as the size of the body or the length of the arms) will
indicate different data values.
Key Points about Stick Figures:
• Data Representation: The data is represented by adjusting different body
parts, such as:
o The size of the body: This could represent the amount of a specific
value.
o The position of the arms: Could show another data point (e.g., a
performance score).
o The length of the legs: Might represent something like growth over
time or any other measurement.
• How it works:
o A stick figure could represent a person’s height, weight, age, or other
measurements.
o Adjusting the size of the body or the proportions of the figure can
show how different categories compare.
• Advantages:
o Stick figures are simple to draw and understand.
o They can show multiple aspects of data in a clear and straightforward
manner.
• Example: Suppose you are comparing the number of sales of three different
products:
o Product A: The stick figure’s arms are spread wide (showing a high
sales number).
o Product B: The arms are at a medium height.
o Product C: The arms are very close to the body (indicating low sales).
This way, you can see the differences in sales visually, just by looking
at how the arms or body sizes change.
How it helps:
 • Simple and Clear: Stick figures are easy to interpret because they resemble
something familiar (a person).
• Good for Comparison: You can quickly compare multiple sets of data just
by looking at the figures side by side.
Summary:
Both Chernoff Faces and Stick Figures are methods of representing data in a way
that is visually intuitive and easy to understand. They allow complex data to be
simplified and presented in a form that most people can quickly grasp, even
without deep knowledge of the numbers involved.
Hierarchical Visualization Techniques
Hierarchical Visualization Techniques are methods used to display data that is
organized in a hierarchy, meaning data is arranged in levels, with higher levels
containing broader categories and lower levels containing more specific details.
These techniques help us to visualize relationships between categories and
subcategories in a way that’s easy to understand.
Here are some hierarchical visualization techniques:

1. Dimensional Stacking:

Dimensional stacking is a method where different levels of data are represented by

stacking them on top of each other, like layers. Each layer represents a different
category or level in the hierarchy.
Key Points:
 • How it works: Imagine a stack of blocks where each block represents a
category. The blocks at the top represent the main categories, and the blocks
at the bottom represent more specific details.
• Example: If you’re looking at a company’s organization, the top block
might represent the company. Below that, you could have blocks for
departments, and even below those, blocks for individual employees.
How it helps:
• Shows Layers: It helps show the different levels of the hierarchy clearly,
making it easier to understand how things are related at different levels.

2. Tree Map:
A tree map is a visual representation of hierarchical data using nested rectangles.
The larger the rectangle, the more significant the category it represents.
Key Points:
• How it works: In a tree map, each rectangle represents a category, and
subcategories are shown as smaller rectangles inside their parent category.
The size of the rectangle can represent the size or importance of the data.
• Example: If you have data about sales in different regions, each region can
be represented as a large rectangle. Inside each region, you can have smaller
rectangles representing individual cities. The bigger the rectangle, the higher
the sales.
How it helps:
• Easy Comparison: You can easily compare the sizes of different categories
by looking at the size of the rectangles.
• Compact: Tree maps show a lot of information in a small space.

3. Cone Trees:
A cone tree is a 3D visualization where the hierarchy is displayed as a cone. The
top of the cone represents the root of the hierarchy, and as you go down, the cone
branches out into smaller sections, showing more specific details.
Key Points:
• How it works: Think of a tree, but instead of branches, the hierarchy is
shown as a 3D cone. The cone tapers down, with each layer representing a
different level of data.
• Example: Imagine you’re looking at a company’s hierarchy. At the top of
the cone, you have the CEO. As you move down, you see the managers,
employees, and departments. The sections widen to show more specific
details.
How it helps:
• 3D View: The cone shape helps represent hierarchical data in 3D, making it
easy to see the relationships between different levels.
• Interactive: You can rotate the cone to see different perspectives and
understand how the levels are connected.
InfoCubes
Steps for Basic Data Visualization in Tableau
Steps for Basic Data Visualization in Tableau
1. Open Tableau and Connect to Data
o Launch Tableau and click "Connect to Data" on the start page.
o Select your data source (e.g., Excel, CSV, or a database) and load
your dataset.
2. Choose the Worksheet
o Navigate to a new worksheet by clicking the Sheet tab at the bottom.
o This is where you'll build your visualization.
3. Drag and Drop Fields
o Drag dimensions (e.g., categories like "Region" or "Product") to the
Columns or Rows shelf.
o Drag measures (e.g., numerical data like "Sales" or "Profit") to the
Columns, Rows, or Marks shelf.
4. Select Chart Type
o Use the "Show Me" panel on the top-right to select a suitable chart
type (e.g., bar chart, line chart, or scatter plot).
o Tableau will suggest chart types based on the data you’ve selected.
5. Enhance the Visualization
o Add colors by dragging a field to the Color shelf in the Marks card.
o Use filters to show specific data by dragging a field to the Filters
shelf.
o Add labels or adjust formatting for better readability.
6. Save or Share
o Save the visualization locally or publish it to Tableau Public or
Tableau Server.
 o Export as an image, PDF, or share it via a link.
By following these steps, you can create simple and effective visualizations in
Tableau.