1

[Link] visualization reference model with diagram.(1,9- in 10 marks & 1,7 in 5

marks)

1. Introduction

What is Visualization?

Visualization is the process of transforming abstract data and information into a visual or
graphical form that humans can easily understand, interpret, and analyze. In simpler terms,
it's about converting numbers, text, and complex datasets into charts, graphs, diagrams, and
other visual representations that communicate patterns, trends, and insights at a glance.

Purpose of a Reference Model

A reference model is a standardized framework or blueprint that describes the structure,

components, and flow of a system. In the context of visualization, a reference model serves
as:

●​ A guide for understanding how visualization systems work

●​ A structured approach to designing effective visualizations
●​ A blueprint for data visualization tools and software
●​ A way to ensure consistency and best practices in visualization design

2. Visualization Reference Model Overview

Concept Definition

The Visualization Reference Model is a conceptual framework that describes the process
of transforming raw data into meaningful visual representations that users can interact with
and gain insights from. It is a systematic pipeline that shows how data flows through multiple
transformation stages before reaching the user's perception.

Historical Background

The Visualization Reference Model was formally proposed by Stuart K. Card, Jock
Mackinlay, and Ben Shneiderman in their seminal work in information visualization. Their
model has become the foundation for understanding how visualization systems function and
is widely used in academic research, industry applications, and educational institutions
worldwide.

Importance in Data Visualization

The significance of this model lies in its ability to:

●​ Bridge the Gap: Connect raw data to human understanding

●​ Guide Design Decisions: Help designers make informed choices at each stage
●​ Enable Tool Development: Provide a framework for building visualization software
●​ Facilitate Communication: Allow visualization experts to discuss and improve
visualization systems systematically

Made With ❤️ By Hari Prasad Korrapolu

 2

●​ Ensure Effectiveness: Help create visualizations that truly communicate insights

rather than just displaying data

3. Neat Diagram of the Visualization Reference Model

4. Detailed Explanation of Each Stage

4.1 Data Transformation

Definition

Data Transformation is the first critical stage where raw, unprocessed data is cleaned,
organized, and prepared for visualization. It involves selecting relevant data, removing
errors, handling missing values, and aggregating or filtering information to create a dataset
suitable for visual analysis.

Process Involved

The data transformation process includes several sub-steps:

Made With ❤️ By Hari Prasad Korrapolu

 3

1.​ Data Collection: Gathering raw data from various sources (databases, sensors,
logs, APIs, spreadsheets)
2.​ Data Cleaning: Removing duplicates, handling missing values, correcting errors,
standardizing formats
3.​ Data Selection: Choosing only the relevant attributes and records for the
visualization
4.​ Data Aggregation: Summarizing data by grouping, averaging, or computing
statistics
5.​ Data Filtering: Removing outliers or irrelevant records that would clutter the
visualization
6.​ Data Integration: Combining data from multiple sources into a unified dataset

Real-World Example: Student Marks Dataset

Scenario: You have raw student data from an examination system containing:

●​ Roll numbers (1-500)

●​ Student names
●​ Subject marks (Mathematics, Science, English, History, Geography)
●​ Attendance percentages
●​ Date of birth
●​ Enrollment date
●​ Extra-curricular activities
●​ Duplicate entries and missing values in some records

Transformation Process:

●​ Cleaning: Remove 15 duplicate records, fill missing attendance values with average
(82%), correct spelling errors in names
●​ Selection: Choose only Roll number, Marks (Math, Science, English), and
Attendance
●​ Aggregation: Calculate average marks per student, classify students into
performance tiers (Excellent: 85+, Good: 70-84, Average: 55-69, Below Average:
<55)
●​ Filtering: Exclude 5 students with incomplete data, focus on students with 75%+
attendance
●​ Result: A clean dataset of 480 students with 5 attributes ready for visualization

4.2 Visual Mapping

Definition

Visual Mapping is the process of converting abstract data values into visible graphical
properties that the human eye can perceive. It's the bridge between numbers and shapes,
where data attributes are mapped to visual channels like color, size, position, and shape.

Process Involved

Made With ❤️ By Hari Prasad Korrapolu

 4

Visual mapping involves:

1.​ Attribute Selection: Choosing which data attributes to visualize

2.​ Channel Assignment: Deciding which visual property represents each data attribute
3.​ Scale Determination: Establishing the range and scale for each visual encoding
4.​ Encoding Rules: Creating a consistent mapping rule (e.g., higher marks = taller bar,
better performance = warmer color)
5.​ Hierarchy Design: Prioritizing which attributes are most important visually
6.​ Aesthetic Selection: Choosing colors, fonts, styles that enhance clarity

Real-World Example: Student Marks Dataset

Mapping Decisions:

●​ X-axis (horizontal position): Student performance tier (4 categories: Excellent,

Good, Average, Below Average)
●​ Y-axis (vertical position): Marks percentage (0-100 scale)
●​ Bar height: Represents the count of students in that category
●​ Color: Represents performance quality (Green for Excellent, Yellow for Good,
Orange for Average, Red for Below Average)
●​ Transparency/Opacity: Represents attendance percentage (more opaque = higher
attendance)

Mapping Example:

●​ Student with 92 marks and 95% attendance → Tall green bar (Excellent),
bright/opaque appearance
●​ Student with 58 marks and 72% attendance → Short orange bar (Average), slightly
transparent

4.3 View Transformation

Definition

View Transformation is the final rendering stage where the visually mapped data is
transformed into a specific projection or coordinate system that displays on the screen. It
handles technical aspects like scaling, positioning, projection, and rendering to create the
final visual output that users see.

Process Involved

View transformation includes:

1.​ Coordinate Transformation: Converting data coordinates to screen coordinates

(pixels)
2.​ Scaling: Adjusting the size of visual elements to fit within the available display area
3.​ Projection: Choosing how 3D data (if applicable) is projected onto 2D screens
4.​ Clipping: Ensuring that visual elements don't extend beyond the display boundaries

Made With ❤️ By Hari Prasad Korrapolu

 5

5.​ Anti-aliasing: Smoothing edges and lines for better visual quality
6.​ Rendering: Drawing all visual elements on the screen with proper colors, strokes,
and fills
7.​ Optimization: Ensuring the visualization displays smoothly and efficiently

Technical Aspects

●​ Aspect Ratio: Maintaining proportional relationships in the visualization

●​ Zoom and Pan: Enabling users to view different parts of the visualization
●​ Resolution Adaptation: Adjusting the visualization for different screen sizes and
devices
●​ Performance Optimization: Ensuring the visualization renders quickly without lag

Real-World Example: Student Marks Dataset

View Transformation Steps:

1.​ Calculate Screen Coordinates:​

○​ Data value range: 0-100 marks

○​ Screen area: 600 pixels wide, 400 pixels tall
○​ Scale factor: 6 pixels per mark on X-axis, 4 pixels per mark on Y-axis
2.​ Position Elements:​

○​ Each bar is drawn at calculated pixel positions

○​ Student with 92 marks → bar height = 92 × 4 = 368 pixels from bottom
○​ All bars are aligned along the bottom axis
3.​ Apply Colors and Styling:​

○​ Green bars for Excellent students

○​ Label and title text positioned relative to bars
○​ Legend placed in a non-overlapping area
4.​ Render Final Output:​

○​ SVG or canvas output displaying the complete chart

○​ All elements visible and properly positioned on the screen
○​ Animation effects applied (if specified)

4.4 Human Interaction

Definition

Human Interaction refers to the feedback loop where users interact with the visualization
through clicking, hovering, filtering, or adjusting parameters. These interactions trigger
changes in earlier stages of the pipeline, creating a dynamic and exploratory visualization
experience.

Made With ❤️ By Hari Prasad Korrapolu

 6

Types of User Interactions

1.​ Selection: Clicking on visual elements to highlight or select them

2.​ Filtering: Using controls to include/exclude specific data subsets
3.​ Zooming: Magnifying specific regions of the visualization
4.​ Panning: Moving the view to explore different areas
5.​ Brushing: Dragging to select a range of data
6.​ Hovering: Displaying tooltips or additional information
7.​ Parameter Adjustment: Using sliders, dropdowns, or text inputs to modify
visualization parameters
8.​ Navigation: Moving between different views or levels of detail

The Feedback Loop

When a user interacts with the visualization:

1.​ The interaction is captured by the interface

2.​ A new query or parameter set is created
3.​ The system returns to earlier stages of the pipeline (data transformation, visual
mapping)
4.​ New visual mappings are computed
5.​ The view is re-rendered with the updated visualization

This creates a tight feedback loop allowing exploratory data analysis.

Real-World Example: Student Marks Dataset

Interactive Scenarios:

1.​ Filtering by Subject:​

○​ User clicks on "Mathematics" in a subject selector

○​ System re-filters data to show only Math marks
○​ Visualization updates to show marks distribution for Math only
2.​ Attendance Filter:​

○​ User moves a slider: "Show students with 80%+ attendance"

○​ System filters the dataset to 350 students (from 500)
○​ Bar chart updates to show only this filtered subset
○​ Chart proportions change accordingly
3.​ Hovering for Details:​

○​ User hovers over a bar

○​ Tooltip appears: "Excellent tier: 120 students, Avg: 88.5%"
○​ Bar highlights in color
4.​ Comparing Performance:​

○​ User selects two subjects to compare side-by-side

○​ System transforms data for both subjects

Made With ❤️ By Hari Prasad Korrapolu

 7

○​ Visualization shows comparative bar charts

○​ Colors differentiate between subjects

5. Pipeline Explanation: How Data Flows Through the

System
The Visualization Reference Model operates as a sequential pipeline with feedback
loops. Let me explain the complete flow:

Step-by-Step Data Flow

Stage 1: Raw Data Input

●​ Data enters the system from various sources (CSV files, databases, APIs, real-time
sensors)
●​ Example: 500 student records with multiple attributes

Stage 2: Data Transformation

●​ Raw data is cleaned and prepared

●​ Example: Removing duplicates → 480 students, Handling missing values,
Standardizing formats, Aggregating into performance tiers

Stage 3: Visual Mapping

●​ Data attributes are assigned visual properties

●​ Example: Marks → Bar height, Performance tier → Color, Attendance → Opacity

Stage 4: View Transformation

●​ Data is converted to screen coordinates

●​ Example: Bar heights calculated in pixels, colors assigned, text positioned, final
SVG/canvas generated

Stage 5: User Visualization and Interaction

●​ User sees the interactive visualization

●​ Example: Sees bar chart with color-coded performance tiers, can filter by attendance
or subject

Stage 6: Feedback Loop

●​ User interacts (e.g., "Show only Science marks")

●​ System captures interaction and loops back to earlier stages
●​ Data transformation re-filters dataset
●​ Visual mapping updates for new subset
●​ View transformation re-renders with new coordinate

Made With ❤️ By Hari Prasad Korrapolu

 8

Describe visual mapping and including techniques, perceptual effectiveness

of visual variables

Visual Mapping in Data Visualization:

1. Introduction to Visual Mapping

Visual Mapping is the critical process of converting abstract data attributes into visible,
perceivable visual properties. It is the bridge between raw numbers and what users actually
see on screen. Without effective visual mapping, even the most interesting data remains
invisible to human understanding.

Made With ❤️ By Hari Prasad Korrapolu

 9

The Core Problem Visual Mapping Solves

Imagine you have a spreadsheet with:

●​ Sales figures: 150,000, 200,000, 175,000

●​ Time periods: January, February, March
●​ Regions: North, South, East

These are abstract numbers and labels. A user cannot directly perceive magnitude or
comparison from these text values. Visual mapping transforms these into:

●​ Heights of bars (position, length)

●​ Colors (hue)
●​ Labels on an axis (position)

Now the user can instantly see that February sales were highest and North is a particular
region with a specific color.

Why Visual Mapping is Critical

The success of any visualization depends entirely on how well data is mapped to visual
channels. Poor mapping leads to:

●​ Misinterpretation of data
●​ Inability to see patterns
●​ Loss of information
●​ Confusion and wasted cognitive effort

2. Understanding Visual Channels

A visual channel is a perceivable property of graphics that can encode data values. The
most important visual channels include:

2.1 Position (Spatial Location)

Definition: Position refers to the location of visual marks on a 2D plane (X and Y

coordinates).

How It Works:

●​ Each data point is placed at a specific (x, y) coordinate

●​ X-axis typically represents one variable
●​ Y-axis represents another variable
●​ Position encodes numerical values with high precision

Example: In a scatter plot of student performance:

●​ X-axis: Hours studied (0 to 10)

●​ Y-axis: Marks obtained (0 to 100)
●​ A student who studied 8 hours and scored 85 marks appears at position (8, 85)

Made With ❤️ By Hari Prasad Korrapolu

 10

Effectiveness: HIGHEST - Position is the most effective visual channel. The human eye can
perceive spatial relationships with extreme precision. Comparing two points' positions is
almost instantaneous and highly accurate.

2.2 Length (Height and Width)

Definition: Length refers to the extent of a visual mark (how tall a bar is, how long a line is).

How It Works:

●​ Data values are mapped to the length of bars, lines, or other shapes
●​ Longer = larger value, Shorter = smaller value
●​ Works best for comparing values within a single dimension

Example: In a bar chart of monthly sales:

●​ Month labels on X-axis (January, February, March)

●​ Each bar's height represents sales amount
●​ Taller bar = higher sales

Effectiveness: VERY HIGH - Length is the second most effective channel. The visual
system rapidly compares lengths, making it ideal for magnitude comparison.

2.3 Direction/Angle

Definition: Direction refers to the angle or orientation of visual marks.

How It Works:

●​ Angles encode directional or categorical information

●​ Used in wind maps (angle shows wind direction)
●​ Used in diverging visualizations (arrows pointing up/down)

Example: In a wind map showing wind direction across a region:

●​ Each location has an arrow

●​ Arrow angle indicates wind direction
●​ Arrow length indicates wind speed

Effectiveness: MODERATE-HIGH - Angle is good for directional data but less precise than
position or length for numerical comparisons.

2.4 Area (2D Size)

Definition: Area refers to the total surface area of a visual mark (like bubble size or shape
area).

How It Works:

●​ Larger area = larger value

●​ Used in bubble charts, treemaps, and area charts

Made With ❤️ By Hari Prasad Korrapolu

 11

●​ Important: human perception of area is non-linear (doubling area doesn't look like
doubling size)

Example: In a bubble chart of company market share:

●​ X-axis: Revenue
●​ Y-axis: Profit
●​ Bubble size: Number of employees
●​ Larger bubbles = more employees

Effectiveness: MODERATE - Area perception is non-linear. People consistently

underestimate size differences. A value that's 4 times larger doesn't appear 4 times bigger.
This makes area less effective than position or length for precise comparisons.

2.5 Color

Color encoding has two primary dimensions:

2.5.1 Hue (Color Identity)

Definition: Hue is the actual color (red, blue, green, yellow).

How It Works:

●​ Different colors represent different categories

●​ Red might represent "North region", Blue "South region"
●​ Used when you want to distinguish groups, not show magnitude

Example: In a stacked bar chart:

●​ Blue bars = North region sales

●​ Red bars = South region sales
●​ Each color instantly identifies its category

Effectiveness: MODERATE - Hue is good for categorical data but poor for showing
ordered/numerical relationships because colors don't have a natural ordering.

2.5.2 Saturation/Brightness (Color Intensity)

Definition: Saturation refers to the intensity or strength of a color. Brightness refers to how
light or dark a color is.

How It Works:

●​ Light color = low value

●​ Dark/intense color = high value
●​ Creates a gradient or spectrum
●​ Used for ordered/continuous data

Example: In a heat map of temperature across regions:

Made With ❤️ By Hari Prasad Korrapolu

 12

●​ Light blue = cold (15°C)

●​ Dark red = hot (35°C)
●​ Intermediate shades show intermediate temperatures

Effectiveness: HIGH - Saturation gradients effectively show ordered data and magnitudes.
Easy for users to perceive "more intense = higher value".

Color Effectiveness Ranking:

1.​ Saturation/Value (for ordered/continuous data) - Most effective

2.​ Hue (for categories) - Good for distinction
3.​ Hue differences (red vs blue) - Moderate without ordering

2.6 Size (Volume/2D Dimension)

Definition: Size refers to the overall visual magnitude of marks, including both length and
width.

How It Works:

●​ Larger symbols = larger values

●​ Used in proportional symbol maps
●​ Important: volume perception is cubic and highly non-linear

Example: In a map showing population by city:

●​ Circle at New York is much larger than circle at smaller cities

●​ Circle size ∝ population

Effectiveness: MODERATE-LOW - Size perception is even more non-linear than area. A

value that's 8 times larger (3D) doesn't appear 8 times bigger. Users systematically
underestimate size differences.

2.7 Texture and Pattern

Definition: Texture and pattern refer to repeated visual elements or fills within marks
(hatching, dots, stripes).

How It Works:

●​ Different patterns represent different categories

●​ Solid fill, hatched fill, striped fill distinguish groups
●​ Useful as secondary encoding alongside color

Example: In a map of land use:

●​ Solid green = forest

●​ Diagonal stripes = urban area
●​ Dots = agricultural area
●​ Helps distinguishing categories for colorblind users

Made With ❤️ By Hari Prasad Korrapolu

 13

Effectiveness: MODERATE - Texture is good for categorical data and helps colorblind
users. Less effective than color for rapid discrimination but more accessible.

2.8 Shape

Definition: Shape refers to the form of visual marks (circles, squares, triangles, diamonds).

How It Works:

●​ Different shapes represent different categories

●​ Circle = one type, Square = another type
●​ Used in scatter plots with multiple series

Example: In a scatter plot of student performance:

●​ Circle markers = science students

●​ Square markers = humanities students
●​ Position (X, Y) shows their scores

Effectiveness: MODERATE - Shape is good for categorical distinctions when color alone
isn't sufficient. Less effective than color for rapid discrimination.

4. Visual Mapping Techniques

4.1 Quantitative Data Mapping

Quantitative data consists of numerical values that have order and magnitude (sales,
temperature, age, income).

Technique 1: Linear Scaling

Concept: Map data values directly to visual channel values with a linear relationship.

Formula:

Visual Value = (Data Value - Min Data) / (Max Data - Min Data) × (Max Visual - Min Visual)

Example: Bar Chart Height:

●​ Data: Sales from $0 to $100,000

●​ Visual: Bar heights from 0 to 400 pixels
●​ Sales of $50,000 → Bar height of 200 pixels

When to Use:

●​ Position (X, Y coordinates)

●​ Length (bar height)
●​ Color saturation (light to dark)

Made With ❤️ By Hari Prasad Korrapolu

 14

Technique 2: Logarithmic Scaling

Concept: Apply logarithmic transformation to data before mapping, useful for data with
exponential ranges.

Formula:

Visual Value = log(Data Value) mapped to visual range

Example: World Population by Country:

●​ Raw data range: 100,000 to 1,400,000,000 (highly skewed)

●​ Log scale: log(100K) ≈ 5.0, log(1.4B) ≈ 9.1 (reasonable range)
●​ Much better visual representation with visible differences

When to Use:

●​ Data spanning many orders of magnitude

●​ When linear scaling makes small values invisible

Technique 3: Color Gradient Mapping (Continuous)

Concept: Map quantitative values to a color gradient from light to dark or across a spectrum.

How It Works:

●​ Minimum value: Light color (e.g., light blue)

●​ Maximum value: Dark color (e.g., dark red)
●​ Intermediate values: Proportional shades in between

When to Use:

●​ Spatial data (maps)

●​ 2D matrices (correlation matrices)
●​ Emphasizing intensity variations

4.2 Categorical Data Mapping

Categorical data consists of discrete groups with no inherent order (regions, product types,
departments).

Technique 1: Color Encoding (Hue)

Concept: Assign distinct colors to different categories.

Implementation:

●​ North region = Blue

●​ South region = Red
●​ East region = Green
●​ West region = Yellow

Made With ❤️ By Hari Prasad Korrapolu

 15

When to Use:

●​ Distinguishing groups
●​ Categorical legends
●​ Segment identification

Design Principles:

1.​ Use perceptually distinct colors: Not similar shades (use blue and red, not light
blue and blue)
2.​ Limit to 5-7 colors: Beyond this, users cannot distinguish all colors
3.​ Consider colorblind users: Avoid red-green combinations
4.​ Use semantic colors: Red for negative/danger, green for positive/success (if
appropriate)

Example - Sales by Region:

Stacked Bar Chart:

Blue segment = North sales

Red segment = South sales

Green segment = East sales

Technique 2: Shape/Symbol Encoding

Concept: Use different shapes to represent different categories.

Implementation:

●​ Product Type A = Circle marker

●​ Product Type B = Square marker
●​ Product Type C = Triangle marker

When to Use:

●​ When you need to encode a third dimension in scatter plot

●​ Color alone is insufficient
●​ Helping colorblind users (shape + color)

Example - Multi-Series Scatter Plot:

Stacked Bar Chart:

Blue segment = North sales

Red segment = South sales

Made With ❤️ By Hari Prasad Korrapolu

 16

Green segment = East sales

Technique 2: Shape/Symbol Encoding

Concept: Use different shapes to represent different categories.

Implementation:

●​ Product Type A = Circle marker

●​ Product Type B = Square marker
●​ Product Type C = Triangle marker

When to Use:

●​ When you need to encode a third dimension in scatter plot

●​ Color alone is insufficient
●​ Helping colorblind users (shape + color)

Example - Multi-Series Scatter Plot:

X-axis: Study hours

Y-axis: Exam score

Shape: Student major

Circle = Science students

Square = Humanities students

Triangle = Commerce students

Technique 3: Pattern/Texture Encoding

Concept: Use patterns (hatching, dots, stripes) to distinguish categories.

Implementation:

●​ Category A = Solid fill

●​ Category B = Diagonal stripes
●​ Category C = Dots
●​ Category D = Horizontal stripes

When to Use:

●​ Grayscale printing requirements

●​ Accessibility for colorblind users
●​ When color cannot be used

Example - Stacked Area Chart:

Made With ❤️ By Hari Prasad Korrapolu

 17

Region 1 = Solid gray

Region 2 = Diagonal stripes

Region 3 = Dots

Region 4 = Horizontal lines

5. Detailed Techniques for Different Data Types

5.1 Univariate Data (One Variable)

Univariate data consists of a single variable measured across many observations.

Example Dataset: Student marks in Mathematics (one subject) for 100 students.

Best Visualization Techniques:

1. Histogram with Position Encoding:

●​ X-axis: Mark ranges (0-10, 10-20, ..., 90-100) — Categorical position

●​ Y-axis: Frequency (count of students) — Length encoding
●​ Height of each bar shows how many students fall in each range
●​ Effectiveness: HIGH - Combines position and length

2. Box Plot:

●​ Position: Median, quartiles shown as vertical lines

●​ Length: Box height shows interquartile range
●​ Whiskers extend to show full data range
●​ Effectiveness: HIGH - Clear distribution view

3. Density Plot with Color Gradient:

●​ X-axis: Marks
●​ Y-axis: Density/frequency
●​ Color gradient: Intensity shows density
●​ Effectiveness: MODERATE-HIGH

5.2 Bivariate Data (Two Variables)

Bivariate data shows relationship between two variables.

Example Dataset: Study hours (X) vs. Exam scores (Y) for 100 students.

Best Visualization Techniques:

1. Scatter Plot (Using Position)

Made With ❤️ By Hari Prasad Korrapolu

 18

2. Bubble Chart (Using Position + Size)

3. 2D Density Heat Map

5.3 Multivariate Data (Three or More Variables)

Multivariate data requires combining multiple visual channels.

Example Dataset: Student performance across 5 subjects.

Discuss visual analytics architecture and component

Made With ❤️ By Hari Prasad Korrapolu

 19

❤️ By Hari Prasad Korrapolu

Made With
 20

Visual Analytics Architecture and Components

1. Introduction

Visual Analytics is the combined use of automated and interactive analytics with
visualization to obtain deep insights from complex, high-dimensional data. It's not just about
creating pretty charts—it's a systematic process where data flows through multiple
processing stages before being presented to users in interactive visual forms.

A Visual Analytics Architecture describes the overall structure, organization, and

interaction of all components required to build an effective visual analytics system.
Understanding this architecture is crucial for:

●​ Designing scalable visualization systems

●​ Optimizing performance for large datasets
●​ Creating seamless user experiences
●​ Implementing real-time analytics

2. The Visual Analytics Pipeline

The complete visual analytics workflow can be divided into five major layers:

2.1 Data Layer: Data Sources and Collection

This is where data originates from multiple sources.

Types of Data Sources:

1. Databases

●​ Relational databases (SQL): PostgreSQL, MySQL, Oracle

●​ NoSQL databases: MongoDB, Cassandra, DynamoDB
●​ Data warehouses: Snowflake, BigQuery, Redshift
●​ Time-series databases: InfluxDB, Prometheus

Example: A retail company's customer transaction database containing millions of purchase

records.

2. APIs and Data Streams

●​ REST APIs from cloud services

●​ Real-time data streams: Kafka, Pub/Sub
●​ WebSocket connections for live updates
●​ Message queues

Example: Stock market data streamed in real-time from financial APIs, updating thousands
of times per second.

3. Files and Logs

Made With ❤️ By Hari Prasad Korrapolu

 21

●​ CSV and Excel files

●​ JSON and XML files
●​ Log files from applications
●​ Archived historical data

Example: Monthly sales CSV files uploaded by regional managers.

4. Sensors and IoT Devices

●​ Temperature, pressure, humidity sensors

●​ GPS and location data
●​ Network traffic monitors
●​ Equipment telemetry

Example: Sensor data from manufacturing equipment monitoring product quality in

real-time.

5. User Input

●​ Form submissions
●​ Manual data entry
●​ File uploads
●​ Parameter selections

Example: Users selecting date ranges or filtering criteria in the analytics dashboard.

2.2 Processing Layer: Data Management and Transformation

Raw data must be prepared and processed before visualization.

Components:

1. Data Integration (ETL)

Definition: Extracting, Transforming, and Loading data from multiple sources into a unified
format.

Tasks:

●​ Extraction: Pulling data from various sources (databases, APIs, files)

●​ Transformation: Cleaning, standardizing, deduplicating, handling missing values
●​ Loading: Storing processed data in a format ready for analysis

2. Data Transformation

Definition: Converting and restructuring data into forms suitable for analysis.

Key Operations:

●​ Aggregation: Summarizing data (sum, average, count)

○​ Example: Total sales by region

Made With ❤️ By Hari Prasad Korrapolu

 22

●​ Filtering: Selecting relevant subsets

○​ Example: Only include sales above $1000
●​ Binning: Grouping continuous values into ranges
○​ Example: Group ages into 10-year bins (0-10, 10-20, 20-30, etc.)
●​ Derivation: Creating new variables from existing ones
○​ Example: Profit margin = (Revenue - Cost) / Revenue
●​ Normalization: Scaling values to a standard range
○​ Example: Convert scores (0-100) to percentages (0-1)

3. Analytics Engine

Definition: Applying computational and statistical methods to extract insights.

Capabilities:

●​ Descriptive Analytics: What happened?

○​ Mean, median, standard deviation
○​ Distribution analysis
○​ Correlation analysis
●​ Diagnostic Analytics: Why did it happen?
○​ Root cause analysis
○​ Trend detection
○​ Anomaly detection
●​ Predictive Analytics: What will happen?
○​ Machine learning models
○​ Forecasting (time-series)
○​ Classification and regression
●​ Prescriptive Analytics: What should we do?
○​ Optimization algorithms
○​ Recommendations
○​ Decision support

2.3 Visualization Layer: Converting Data to Graphics

The processed data is converted into visual representations.

Components:

1. Visual Mapping

Definition: Assigning data attributes to visual properties (position, color, size, shape).

Process:

●​ Quantitative data (marks 0-100) → Visual channel (bar height 0-400 pixels)
●​ Categorical data (regions: North, South, East) → Visual channel (colors: blue, red,
green)
●​ Temporal data (months) → Visual channel (X-axis position)

Made With ❤️ By Hari Prasad Korrapolu

 23

2. Chart Rendering

Definition: Actually drawing the visual representations on screen.

Rendering Technologies:

●​ SVG (Scalable Vector Graphics)

○​ Good for: Interactive charts, diagrams
○​ Advantage: Vector-based (scales perfectly)
○​ Limitation: Can be slow with thousands of elements
●​ Canvas (HTML5 Canvas)
○​ Good for: Pixel-based graphics, animations
○​ Advantage: Fast rendering of complex graphics
○​ Limitation: Not easily interactive at element level
●​ WebGL (Web Graphics Library)
○​ Good for: 3D visualizations, massive datasets
○​ Advantage: GPU acceleration, renders millions of points
○​ Limitation: Steep learning curve
●​ GPU-Accelerated Libraries
○​ [Link]: Large-scale geospatial visualizations
○​ [Link]: 3D visualizations
○​ Mapbox GL: Interactive maps

3. Layout Engine

Definition: Positioning visual elements correctly in the available space.

Responsibilities:

●​ Coordinate Transformation: Converting data values to screen pixels

○​ Data range: 0-100 marks
○​ Screen range: 0-400 pixels
○​ Formula: pixel_pos = (data_value - min_data) / (max_data - min_data) ×
screen_width
●​ Scaling: Adjusting size to fit viewport
○​ Linear scale: proportional mapping
○​ Logarithmic scale: for exponential ranges
○​ Power scale: for non-linear relationships
●​ Projection: Converting 3D or geographic data to 2D
○​ Mercator projection for maps
○​ Orthographic projection for 3D views
●​ Zoom and Pan: Managing view transformations
○​ Zoom in: increase scale
○​ Pan: translate coordinates

2.4 Presentation and Interaction Layer

Users interact with visualizations to explore data and gain insights.

Made With ❤️ By Hari Prasad Korrapolu

 24

Components:

1. Interactive Visualization

Definition: Visualizations that respond to user actions in real-time.

Interaction Techniques:

●​ Selection: Choosing specific data points or groups

○​ Click on a bar to highlight it
○​ Drag to select a region
○​ Ctrl+Click for multiple selection
●​ Filtering: Showing only relevant data
○​ Slider to filter by date range
○​ Checkbox to include/exclude categories
○​ Text input for keyword filtering
●​ Sorting: Reordering displayed data
○​ Sort bars by value (ascending/descending)
○​ Sort by alphabetical order
○​ Custom sort by attributes
●​ Drilling Down: Exploring data at finer granularity
○​ Click on "North Region" to see individual branches
○​ Click on "January" to see daily breakdown
○​ Expand tree node to show children
●​ Zooming: Magnifying or shrinking view
○​ Zoom in to see details
○​ Zoom out for overview
○​ Semantic zoom (change level of detail)
●​ Panning: Moving around large visualizations
○​ Drag to move the view
○​ Scroll wheel to pan
○​ Arrow keys for navigation
●​ Hovering: Revealing additional information
○​ Tooltip showing exact values
○​ Highlight related items
○​ Preview of related data

2. User Interface Controls

Definition: Widgets that allow users to adjust parameters and visualizations.

Control Types:

●​ Sliders
○​ Range selection (start date to end date)
○​ Single value (year selection)
○​ Multi-slider (price range)
●​ Dropdowns/Selectors
○​ Choose metric (Revenue, Profit, Cost)

Made With ❤️ By Hari Prasad Korrapolu

 25

○​ Select visualization type (Bar, Line, Pie)

○​ Pick color scheme
●​ Buttons
○​ Toggle (show/hide series)
○​ Action (export, print, share)
○​ Navigation (next, previous)
●​ Checkboxes
○​ Include/exclude categories
○​ Show/hide layers
○​ Apply filters
●​ Color Pickers
○​ Custom color selection
○​ Theme selection
○​ Highlight color
●​ Input Fields
○​ Search text
○​ Numeric thresholds
○​ File upload

3. Core Components and Their Functions

3.1 Data Store

Purpose: Efficiently storing and accessing data during analytics session.

Implementation:

●​ In-Memory Cache: Keeps frequently accessed data in RAM for speed

●​ Indexing: Creates data structures for fast lookups
○​ B-tree indices for range queries
○​ Hash indices for equality queries
○​ Spatial indices for geographic data
●​ Partitioning: Divides data into manageable chunks
○​ Vertical partitioning: separate columns
○​ Horizontal partitioning: separate rows by value

3.2 Query Engine

Purpose: Efficiently extracting relevant subsets of data based on user queries.

Purpose: Efficiently extracting relevant subsets of data based on user queries.

Operations: 1. Filtering 2. Aggregation 3. Sorting4. Grouping 5. Joining

3.3 Graphics System

Purpose: Rendering visual elements on screen.

Components:

Made With ❤️ By Hari Prasad Korrapolu

 26

1. Coordinate System

●​ Maps data values to screen pixels

●​ Defines axes and scales
●​ Handles viewport transformations

2. Graphic Marks

●​ Rectangles (bars)
●​ Circles (bubble charts)
●​ Lines (line charts)
●​ Paths (curves, polygons)

3. Scales

●​ Linear scale: uniform mapping

●​ Log scale: exponential data
●​ Time scale: temporal data
●​ Ordinal scale: categorical data

4. Axes

●​ X-axis, Y-axis (2D)

●​ Z-axis (3D)
●​ Radial axis (polar plots)

5. Legends

●​ Color legend
●​ Shape legend
●​ Size legend

3.4 Event System

Purpose: Handling all user interactions and system events.

Event Types:

1. Mouse Events

●​ Click: select, drill-down

●​ Double-click: expand details
●​ Hover: show tooltip
●​ Drag: select range, move objects
●​ Scroll: zoom or pan

2. Keyboard Events

●​ Arrow keys: navigate

●​ Delete key: remove selection

Made With ❤️ By Hari Prasad Korrapolu

 27

●​ Ctrl+Z: undo
●​ Ctrl+S: save

3. Touch Events (for mobile/tablet)

●​ Tap: select
●​ Long-press: context menu
●​ Pinch: zoom
●​ Swipe: pan or navigate

4. Custom Events

●​ Filter changed
●​ Data updated
●​ Visualization switched
●​ Parameter adjusted

Note:Till now we cover

visualization ref model ,visual
mapping ,visual analytics ,and
we missed only design of
visualization applications
using the above info you can
write

Made With ❤️ By Hari Prasad Korrapolu

 28

❤️ By Hari Prasad Korrapolu

Made With