DATA VISUALIZATION & ANALYTICS

Comprehensive Exam-Ready Notes

Covering: Data Storytelling | Visualization Principles | EDA | Legal Analytic
UNIT 1: DATA STORYTELLING & NARRATIVE VISUALIZATION

1.1 What is Data Storytelling?

Definition: Data storytelling is the structured communication of insights derived from data
using visualizations and contextual explanation. It converts analytical findings into
understandable insights.

Data storytelling sits at the intersection of four elements (Lydia Hooper's model):
• Visual Design - Design principles, design elements, imagery
• Context for Communication - Understanding the audience, clear purpose/goal, feedback loops
• Data - Quality data source, statistical analysis, accurate representation
• Narrative - Language & messaging, beginning, middle & end, call to action

1.2 Data Storytelling vs. Data Analysis vs. Data Communication

Aspect Data Analysis Data Communication
Focus Calculations, statistical techniques, Explaining insights clearly, structuring
pattern identification visuals logically
Output Patterns, statistical findings Structured visuals guiding audience to
understanding
Goal Find what the data shows Communicate why it matters and what
decisions to make

1.3 Three Fundamental Questions Data Storytelling Answers

• What does the data show?
• Why is it happening?
• What decision can be made from it?

1.4 Conceptual Flow / Pipeline

The logical sequence in any data project:
Step Activity
1 Data Collection
2 Data Storage (Database Systems)
3 Data Cleaning
4 Data Visualization
5 Insight Generation
6 Storytelling and Reporting
1.5 Importance Across Domains
• Corporate decision-making
• Policy analysis
• Risk assessment
• Financial reporting
• Cyber security analytics

Cyber Security Example: Instead of raw log entries, a dashboard shows: number of attacks
per day, source distribution, and severity trends.

1.6 Narrative Visualization

Definition: Narrative visualization organizes multiple visuals in a structured sequence to
guide the audience's interpretation — unlike isolated charts.

Standard narrative sequence:

• Overview → Pattern Identification → Comparison → Conclusion
Two narrative approaches:
Approach Description
Author-Driven Presenter controls the flow (like a slideshow or report)
Reader-Driven Audience interacts with the dashboard freely

1.7 Popular Data Visualization Tools

Tool Type Key Feature
Tableau Desktop/Web Advanced interactive analytics
Power BI Microsoft Enterprise analytics, Excel integration
Excel Desktop Basic visualization, widely accessible
Google Looker Studio Web-based Free, web-based reporting & dashboards

1.8 Interactive Visualization

Interactive visualization allows dynamic, user-driven exploration of data.
Core components:
• Filters - Narrow data by category, date, region, etc.
• Drill-down features - Click to explore sub-categories
• Tooltips - Hover to see exact values
 • Dynamic updates - One filter selection automatically updates all related charts

Example: Selecting a year in a dashboard automatically updates all charts on that page.
UNIT 2: CREATING CHARTS & GRAPHS

2.1 Chart Selection Guide (Very Important for Exams)

Data Type Best Chart Example Use
Categorical Bar Chart Sales by region, crime type distribution
Time-based (trends) Line Chart Monthly revenue, conviction rates over years
Continuous Histogram Employee experience levels, age distribution
(distribution)
Proportion Pie Chart Market share (use sparingly)
Relationship (two Scatter Plot Study hours vs. marks, literacy vs. crime rate
variables)

2.2 Bar Chart

• Purpose: Comparing categories
• How to read: Horizontal axis = categories; Vertical axis = values; taller bar = higher value
• Variations: Grouped bar chart (multiple categories side by side), Stacked bar chart

Exam Tip: Use bar charts when you need to compare distinct groups. Best for categorical
data.

2.3 Line Chart

• Purpose: Identifying trends over time
• How to read: Horizontal axis = time; Vertical axis = measurable quantity; slope = direction of
change
• Upward slope = increase; Downward slope = decrease; Flat = no change

Exam Tip: Use line charts ONLY for time-based data. Multiple lines allow comparison of
multiple categories over time.

2.4 Histogram
• Purpose: Understanding distribution of continuous numerical variables
• How to read: Each bar represents a range (interval); height = frequency within that range
• Unlike bar charts, bars in histograms are TOUCHING (no gaps) because data is continuous
Histogram Distribution Types:
Distribution Shape Interpretation
Normal Bell-shaped, symmetric Data evenly distributed around the center
Right Skew (Positive) Tail extends to the right Most values are low; few are very high (e.g.,
salary data)
 Left Skew (Negative) Tail extends to the left Most values are high; few are very low (e.g.,
easy exam marks)
Uniform Flat, no peaks Equal spread across all intervals
Bimodal Two peaks Two distinct groups in the data

Remember: Skew = where the TAIL is. Tail on right = Right skew. Tail on left = Left skew.
No tail = Symmetric.

2.5 Scatter Plot

• Purpose: Examining relationships / associations between two numerical variables
• How to read: Each point = one observation; axis X = one variable; axis Y = another variable
• Pattern of points reveals the relationship
Pattern Meaning
Points rising left to right Positive correlation (both variables increase together)
Points falling left to right Negative correlation (one increases, other decreases)
No clear pattern No correlation (no relationship between variables)
Tight cluster around a line Strong correlation
Scattered widely Weak correlation

Critical Warning: Correlation does NOT mean causation! A pattern in a scatter plot shows
association, not cause-and-effect. Always require additional evidence for causation.

2.6 Box Plot (Advanced)

• Purpose: Showing distribution, spread, and outliers of numerical data
Components of a Box Plot:
• Central line: Median (middle value)
• Box: Contains middle 50% of data (IQR - Interquartile Range)
• Whiskers: Extend to minimum and maximum within 1.5x IQR
• Points outside whiskers: Outliers (unusual extreme values)

Interpretation: Longer box = greater variation in data. Distant outliers = unusual/extreme

cases that need investigation.

Legal Application: Comparing sentencing durations across different courts, or case

pendency levels across states.

2.7 Heatmap (Advanced)

• Purpose: Showing relationships among MULTIPLE variables simultaneously
• How to read: Color intensity represents the strength of the relationship
Color/Shade Meaning
Darker shade Stronger relationship between the two variables
Lighter shade Weaker relationship
Positive value Variables move in the same direction
Negative value Variables move in opposite directions
UNIT 3: PRINCIPLES OF EFFECTIVE DATA VISUALIZATION

3.1 Why Visual Perception Matters

Data visualization is effective because the human brain evolved to process visual information
efficiently — not because charts look nice.
Key cognitive facts:
• Visual information is processed FASTER than text or numbers
• Patterns and differences are detected almost immediately
• Anomalies draw attention without conscious effort
• Viewers form impressions BEFORE detailed reasoning begins

Implication: Poorly designed visuals can mislead viewers without them realizing it. This is
especially dangerous in legal and policy contexts.

3.2 Visual Perception vs. Visual Cognition

Concept Definition In Practice
Visual Perception The brain receives and organizes What the viewer notices first; which
visual information elements attract attention
Visual Cognition Interpreting, reasoning, and Drawing inferences; understanding what
retaining visual patterns the pattern means

A successful visualization supports BOTH processes simultaneously.

3.3 The Five Core Principles

Principle 1: Clarity Over Decoration
• The #1 purpose of any visualization is clear communication
• Elements that REDUCE clarity: decorative backgrounds, excessive gridlines, ornamental
icons, 3D effects
• From a cognitive perspective: decorative elements increase mental effort — viewer wastes
energy distinguishing data from visual noise

Rule: If a design element does not convey data, remove it.

Principle 2: Use Position and Length for Comparison

• Human perception is MOST accurate when comparing position along a common scale and
length
• This is why bar charts and line charts are among the most effective chart types
• Bar charts: eye compares bar heights quickly and accurately
• Line charts: eye naturally follows direction of a line; trends become immediately apparent
Principle 3: Avoid Cognitive Overload
• Human working memory has limited capacity
 • Cognitive overload occurs when a chart has: too many colors, multiple overlapping variables,
dense labels or annotations
Best practices to avoid overload:
• One key idea per chart
• Break complex information into multiple simpler visuals
• Guide interpretation step by step (similar to structured legal reasoning)
Principle 4: Use Colour with Purpose
• The human eye detects color differences rapidly and assigns meaning even when none is
intended
Appropriate uses of color:
◦ Highlighting important categories
◦ Distinguishing between groups
◦ Representing intensity or magnitude
Inappropriate uses of color:
◦ Arbitrary or decorative color choices
◦ Excessive color variation
◦ Combinations that are difficult for color-blind viewers to distinguish
Rule: Color should SUPPORT interpretation, not dominate it.

Principle 5: Maintain Honest Scales and Proportions

• Viewers trust visual proportions instinctively and may not check axis labels carefully
• Misleading practices to AVOID: truncated axes, uneven scales, exaggerated visual
differences

Ethical Rule: Never manipulate scales to exaggerate or conceal facts — especially in legal,
policy, and governance contexts where decisions affect real people.

3.4 Pre-Attentive Processing

Definition: Certain visual features are detected almost instantaneously, BEFORE conscious
effort. These include: position, size, color, orientation, and movement.

Effective visualizations intentionally use these pre-attentive features to direct attention:

• A differently colored bar among uniform bars attracts immediate attention
• A significantly larger element stands out without deliberate analysis

3.5 Gestalt Principles of Visual Organization

Gestalt psychology explains how humans naturally group visual elements:
Principle Meaning Application in Charts
Proximity Elements placed close together Group related data points near each other
are perceived as related
Similarity Elements with similar Use the same color for the same category
appearance are grouped
 together
Continuity The eye prefers smooth, Why line charts are easy to follow
continuous paths
Enclosure Elements within boundaries are Use bordered boxes to group related charts
seen as belonging together

Key Point: Proper grouping reduces mental effort. Poor grouping leads to confusion and
misinterpretation.
UNIT 4: EXPLORATORY DATA ANALYSIS (EDA)

4.1 What is EDA?

Definition: EDA (Exploratory Data Analysis) is the FIRST stage of analysis undertaken after
data collection and BEFORE drawing conclusions or inferences. It is the process of
understanding the nature and structure of the dataset.

Key objectives of EDA:

• Understanding the nature and structure of the dataset
• Assessing data reliability and completeness
• Identifying unusual or extreme values (outliers)
• Discovering potential relationships between variables

Critical Note: EDA does NOT establish guilt, liability, or causation. Its role is LIMITED to
understanding factual patterns reflected in data.

4.2 EDA and Visualization: Inseparable Partners

• EDA is a process of exploration — visualization is its primary instrument
• Whenever we draw graphs to spot trends, compare categories with bars, or check
relationships visually, we are performing EDA
• Meaningful exploratory analysis is virtually impossible without effective visualization

Legal Analogy: EDA is like carefully reading the facts of a case before applying legal
provisions. Just as lawyers must understand facts before arguing law, data analysts must
explore data before drawing conclusions.

4.3 Understanding Data Types

Data Type Description Legal Examples Suitable Charts
Categorical Classifies Type of offence, gender, case Bar chart, pie chart
information into category (civil/criminal)
distinct groups
Numerical Measured values Age of accused, fine amount, Histogram, box plot,
(Continuous) that can take any case duration scatter plot
value in a range
Numerical Values recorded at Conviction rates by year, Line chart
(Time-based) different points in monthly case filings
time

4.4 The Five-Step EDA Process

Step Name What to Do
 1 Understand the What does each variable mean? What units? Are there missing
Dataset values?
2 Data Cleaning Identify missing entries, duplicated records, logically impossible
values
3 Univariate Analysis Examine ONE variable at a time (histograms, box plots)
4 Bivariate Analysis Examine TWO variables together (scatter plots, grouped bar charts)
5 Multivariate Analysis Examine MULTIPLE variables simultaneously to understand
complexity

Step 3: Univariate Analysis

• Questions to ask: What is the typical value? How spread is the data? Are extreme values
present?
• Tools: Histograms (for distribution shape), box plots (for spread and outliers)
Step 4: Bivariate Analysis
• Questions to ask: Is there a visible relationship? Do changes in one variable correspond with
changes in another?
• Tools: Scatter plots (for association), grouped bar charts (for category comparison)
Step 5: Multivariate Analysis
• Purpose: Understand complexity, observe combined effects, generate further questions
• Tools: Heatmaps (correlation matrices), multiple regression plots

4.5 Correlation vs. Causation (Crucial Distinction)

Concept Meaning Example
Correlation Two variables are statistically Ice cream sales and drowning rates both
associated / move together rise in summer
Causation One variable directly CAUSES the Smoking causes lung cancer (proven
other to change through controlled studies)

Exam Rule: A scatter plot showing a pattern ONLY proves correlation, NOT causation.
Causation requires additional controlled research and evidence.
UNIT 5: LEGAL ANALYTICS & ETHICAL VISUALIZATION

5.1 Why Visualization Matters in Legal Contexts

• Legal work increasingly involves large volumes of data: crime statistics, case pendency
figures, conviction rates, compliance reports, financial disclosures
• Tabular data is difficult to compare, time-consuming, and susceptible to misreading
Visualization converts this into:
◦ Clearer patterns — immediate identification of trends tables obscure
◦ Faster comparisons — quicker, more reliable comparisons across datasets
◦ Immediate recognition — instant trend recognition, similar to how flowcharts simplify legal
procedures

Key Point: Judges, policymakers, and administrators are also human viewers. Visuals
influence their interpretation subconsciously.

5.2 Legal Applications of Chart Types

Chart Type Legal Application
Bar Chart Comparing case volumes across different offense categories, jurisdictions, or
time periods
Line Chart Tracking case pendency rates, conviction trends, or crime statistics across
multiple years
Histogram Understanding distribution of sentence lengths or case durations
Scatter Plot Examining relationships between variables like literacy rates and crime rates
Box Plot Comparing sentencing durations across courts or pendency levels across
states
Heatmap Showing correlations among multiple legal variables simultaneously

5.3 Ethical Visualization: Avoiding Manipulation

Rule: Visualizations must NEVER exaggerate or conceal facts. In legal and governance
contexts, misleading visualization can result in unfair or incorrect decision-making.

Three common forms of manipulative visualization:

Manipulation Type How It Works Why It's Dangerous
Manipulated/Truncated Y-axis doesn't start at 0; small Viewers instinctively trust visual
Scales differences look huge proportions without checking axes
Incomplete Datasets Selective data that hides important Can create false narratives that
context favor one argument
Cherry-Picked Periods Selective time ranges that distort Makes a declining trend look like
trends growth, or vice versa

5.4 Common Errors to Avoid

Error Description Correct Approach
Treating graphs as Using visuals just to make slides Every visualization must answer a
decoration attractive, without analytical purpose specific analytical question
Ignoring scales and Drawing conclusions without Always verify axis labels and units
units checking what axes represent and before interpreting
their units
Drawing Interpreting data without legal, Data patterns require contextual
conclusions without social, and historical context interpretation
context
Correlation implies Assuming a visual relationship Association must be verified through
causation means one causes the other additional controlled evidence
Cognitive overload Too many colors, variables, or labels One chart should answer one clear
in one chart analytical question
Excessive color Using too many colors arbitrarily Use color purposefully — to highlight or
use distinguish, not decorate

5.5 Responsible Visualization Checklist

• Is the visualization accurate and honest?
• Does it avoid emotional manipulation?
• Does it support reasoning without replacing it?
• Is the scale honest (not truncated or exaggerated)?
• Is color used purposefully?
• Is cognitive load minimal (one chart, one idea)?
• Is the chart type appropriate for the data type?
QUICK REVISION: KEY DEFINITIONS
Term Definition
Data Storytelling Structured communication of data-derived insights using visualizations
and narrative context
Narrative Visualization Organizing multiple visuals in a sequence to guide interpretation
toward a conclusion
EDA (Exploratory Data First-stage systematic examination of a dataset before drawing
Analysis) conclusions
Cognitive Load The mental effort required to process information; effective
visualization minimizes this
Pre-Attentive Processing Detection of visual features (color, size, position) instantly before
conscious thought
Gestalt Principles How humans naturally group visual elements: proximity, similarity,
continuity, enclosure
Correlation Statistical association between two variables (they move together or
opposite)
Causation One variable directly causes change in another (requires controlled
evidence)
Skew Asymmetry in a data distribution; right skew = tail to right; left skew =
tail to left
Outlier An extreme value that falls far outside the typical range of the dataset
Univariate Analysis Examination of one variable at a time (histograms, box plots)
Bivariate Analysis Examination of two variables together (scatter plots, grouped bar
charts)
Multivariate Analysis Simultaneous examination of three or more variables (heatmaps)
Author-Driven Narrative Presenter controls the visualization flow (reports, slideshows)
Reader-Driven Narrative Audience explores data interactively (dashboards)

Final Exam Tip: Remember the key rule: Effective visualization is NOT defined by software
tools, but by how well it supports human understanding. Clarity, accuracy, and context are
ALWAYS paramount.