Week 5

From Analysis to Argument

Transforming Data Insights into Compelling Visualizations

DATA6000: Industry Business Analytics Project

Use ← → arrow keys or buttons to navigate

Today's Learning Objectives

By the end of this workshop, you will be able to:

Select appropriate visualizations by matching data types to chart types using the VINE framework
Articulate visualization intent — knowing what story each chart should tell
Evaluate visualization effectiveness using the "evidence test"
Plan your Assessment 3 methodology using the MAP framework

            Where You Are: You've submitted Assessment 1 with descriptive analysis. Now we bridge to Assessment 3's predictive analytics.
        

The Visualization Selection Problem

Students often create visualizations that don't support their argument. Why?

✗ Common Mistakes

Pie chart with 15 categories
Bar chart for time trends
3D effects that distort data
Title describes chart, not insight

✓ What We Want

Chart type matches the data
Chart type matches the intent
Visual clearly supports the claim
Title states the insight

The Solution: A systematic framework to match your data, your intent, and your chart selection.

1

The VINE Framework

Variables → Intent → Narrative → Evidence

Introducing the VINE Framework

A 4-step systematic approach for creating effective visualizations:

V

Variables

"What type of data am I showing?"

I

Intent

"What claim am I making?"

N

Narrative

"Which chart tells this story?"

E

Evidence

"Does my visual prove my point?"

Key Insight: Many visualization failures happen because students skip straight to "N" (picking a chart) without considering Variables and Intent first.

V — Variables: Know Your Data Types

The first step is identifying what type of data you're working with:

Data Type	Definition	Examples
Numerical Continuous	Can take any value within a range	Sales ($), Temperature, Duration
Numerical Discrete	Countable, whole numbers only	Number of orders, Headcount, Ratings (1-5)
Categorical Nominal	Categories with no natural order	Product type, City, Payment method
Categorical Ordinal	Categories with a meaningful order	Education level, Satisfaction (Low/Med/High)
Time/Date	Temporal data	Order date, Month, Year, Quarter

Quiz: Identify the Variable Type

Q1: "Customer Segment" with values: Consumer, Corporate, Home Office

A) Numerical Continuous

B) Numerical Discrete

C) Categorical Nominal

D) Categorical Ordinal

✓ Correct! These are categories with no inherent order.

✗ Not quite. These are categories (Consumer, Corporate, Home Office) with no inherent order, making it Categorical Nominal.

Quiz: Identify the Variable Type

Q2: "Monthly Revenue" measured in dollars over 24 months

A) Numerical Continuous + Time

B) Categorical Nominal

C) Numerical Discrete only

D) Categorical Ordinal

✓ Correct! Revenue is continuous, and it's measured over time.

✗ Revenue in dollars is continuous (can be any value), and "monthly" indicates a time dimension.

Quiz: Identify the Variable Type

Q3: "Education Level" with values: High School, Bachelor's, Master's, PhD

A) Categorical Nominal

B) Categorical Ordinal

C) Numerical Discrete

D) Time/Date

✓ Correct! These categories have a natural progression/order.

✗ These are categories with a meaningful order (HS → Bachelor's → Master's → PhD), making it Categorical Ordinal.

I — Intent: What Story Are You Telling?

Every visualization should have a clear purpose. There are 5 main intents:

Intent	You Want To Show...	Example Question
Comparison	Differences between groups	"Which region has highest sales?"
Composition	Parts of a whole	"What % of revenue from each product?"
Distribution	How values are spread	"How are customer ages distributed?"
Relationship	Connection between variables	"Does price affect quantity sold?"
Trend	Change over time	"How have sales changed over 3 years?"

            SPAR Connection: Patterns suggest Comparison or Distribution. Anomalies suggest Trend. Relationships suggest... Relationship!
        

Quiz: What's the Intent?

Q4: You found that Electronics has 35% higher profit margin than Furniture. What intent best represents this finding?

A) Comparison

B) Composition

C) Distribution

D) Trend

✓ Correct! You're comparing profit margins between two categories.

✗ You're comparing two categories (Electronics vs Furniture), so the intent is Comparison.

Quiz: What's the Intent?

Q5: You discovered that Q4 consistently shows a 40% spike in sales every year for the past 5 years.

A) Comparison

B) Composition

C) Distribution

D) Trend

✓ Correct! You're showing change over time with a seasonal pattern.

✗ This is about change over time (5 years, Q4 patterns), making the intent Trend.

Quiz: What's the Intent?

Q6: You want to show that Technology contributes 45%, Office Supplies 35%, and Furniture 20% of total revenue.

A) Comparison

B) Composition

C) Relationship

D) Distribution

✓ Correct! You're showing parts of a whole (percentages that sum to 100%).

✗ The percentages sum to 100% — you're showing parts of a whole, which is Composition.

Quiz: What's the Intent?

Q7: You found a correlation of 0.78 between marketing spend and sales revenue.

A) Comparison

B) Composition

C) Relationship

D) Trend

✓ Correct! Correlation shows how two variables relate to each other.

✗ A correlation measures the connection between two variables — this is Relationship.

N — Narrative: The Chart Decision Matrix

Combine your Variable type and Intent to select the right chart:

Intent	Categorical Data	Numerical Data	Time Data
Comparison	Bar Chart	Bar Chart, Dot Plot	Line Chart (multiple)
Composition	Stacked Bar, Pie (≤5)	Treemap	Stacked Area
Distribution	Bar Chart (counts)	Histogram, Box Plot	—
Relationship	Heatmap	Scatter Plot	Connected Scatter
Trend	—	Line Chart	Line Chart, Area

Common Error: Using pie charts for more than 5 categories or for comparison — use bar charts instead!

Chart Quick Reference

📊

Bar Chart

Compare categories

📈

Line Chart

Show trends over time

🥧

Pie Chart

Parts of whole (≤5)

⬡

Scatter Plot

Show relationships

📉

Histogram

Show distribution

📦

Box Plot

Compare distributions

🔥

Heatmap

Show intensity/correlation

📐

Stacked Bar

Composition + comparison

Quiz: Which Chart?

Q8: You have monthly sales data for 3 years and want to show the overall trend with seasonal patterns. Which chart?

A) Pie Chart

B) Line Chart

C) Bar Chart

D) Scatter Plot

✓ Correct! Line charts are best for showing trends over time.

✗ For time-based trends with seasonal patterns, a Line Chart is the best choice.

Quiz: Which Chart?

Q9: You want to compare average order value across 8 different product categories. Which chart?

A) Pie Chart

B) Line Chart

C) Bar Chart

D) Histogram

✓ Correct! Bar charts are ideal for comparing categories.

✗ For comparing values across categories, use a Bar Chart. (Pie charts shouldn't have 8 categories!)

Quiz: Which Chart?

Q10: You want to investigate if there's a correlation between customer satisfaction score and purchase frequency.

A) Bar Chart

B) Pie Chart

C) Scatter Plot

D) Line Chart

✓ Correct! Scatter plots show the relationship between two numerical variables.

✗ To show the relationship between two numerical variables, use a Scatter Plot.

Quiz: Which Chart?

Q11: You want to show that Technology (45%), Office Supplies (35%), and Furniture (20%) make up total revenue.

A) Pie Chart

B) Scatter Plot

C) Line Chart

D) Histogram

✓ Correct! Pie charts work for composition with ≤5 categories.

✗ With only 3 categories showing parts of a whole, a Pie Chart is appropriate.

E — Evidence: The Visualization Tests

Before finalizing your visualization, apply these three tests:

1. The 3-Second Test

Can someone understand your main message within 3 seconds?

If not → simplify or change chart type

2. The Title Test

Does your title state the insight, not just describe the chart?

3. The Evidence Test

Does the visual clearly support your claim?

Ask: "If someone disagreed, does this chart prove them wrong?"

✗ Descriptive Title

"Sales by Region"

"Monthly Revenue Chart"

✓ Insight Title

"West Region Leads with 40% of Total Sales"

"Revenue Grew 25% in Q4 2024"

Quiz: Good Title or Bad Title?

Q12: "Customer Satisfaction by Department"

A) Bad — Only describes what's shown, no insight

B) Good — Clear and professional

✓ Correct! This title describes but doesn't tell us the finding.

✗ This title only describes what's shown. A better title would state the insight, like "IT Department Has Lowest Satisfaction."

Quiz: Good Title or Bad Title?

Q13: "IT Department Has Lowest Satisfaction Score at 3.2/5"

A) Bad — Too specific

B) Good — States the key insight with evidence

✓ Correct! This title makes a claim and provides supporting data.

✗ This is actually a good title — it states the insight (IT lowest) with evidence (3.2/5).

Quiz: Good Title or Bad Title?

Q14: "Revenue Trend 2020-2024"

A) Bad — Describes the chart, not the finding

B) Good — Clear time range specified

✓ Correct! This describes what we see, not what we learned.

✗ Specifying a time range isn't an insight. What happened during 2020-2024? Did it go up or down?

Quiz: Good Title or Bad Title?

Q15: "Revenue Doubled from $2M to $4M Between 2020-2024"

A) Bad — Too much detail in title

B) Good — Clear insight with specific evidence

✓ Correct! This tells us exactly what happened with supporting numbers.

✗ The specific numbers strengthen the insight. This is an excellent title.

VINE Framework Summary

V

Variables

Identify: Numerical, Categorical, Time

I

Intent

Choose: Compare, Compose, Distribute, Relate, Trend

N

Narrative

Select chart using Decision Matrix

E

Evidence

Test: 3-Second, Title, Evidence

Remember: Follow VINE in order! Don't jump to chart selection (N) before understanding your Variables (V) and Intent (I).

2

From Descriptive to Predictive

Planning Your Assessment 3 Methodology

The Analytics Progression

Assessment 1 focused on descriptive analytics. Assessment 3 requires you to go further:

Descriptive

"What happened?"

✓ Assessment 1

Diagnostic

"Why did it happen?"

Bridge

Predictive

"What will happen?"

Target

Prescriptive

"What should we do?"

Target

            Your Task: Transform your Assessment 1 findings into predictive questions for Assessment 3.
        

Example: From Descriptive to Predictive

Level	Example Finding/Question
Descriptive (A1 finding)	"West region has 23% higher customer churn than East region"
Diagnostic (Why?)	"West region has longer delivery times and lower satisfaction scores"
Predictive (A3 question)	"Can we predict which customers will churn based on delivery time and satisfaction?"
Prescriptive (A3 recommendation)	"Target customers with >5 day delivery and <3 satisfaction with retention offers"

Notice: Each level builds on the previous. Your descriptive findings become the foundation for predictive models.

Quiz: Identify the Analytics Level

Q16: "Sales increased by 15% in Q4 compared to Q3"

A) Descriptive

B) Diagnostic

C) Predictive

D) Prescriptive

✓ Correct! This describes what happened in the past.

✗ This statement describes what happened (past facts) — that's Descriptive analytics.

Quiz: Identify the Analytics Level

Q17: "Based on current trends, we forecast Q1 sales to reach $2.3M"

A) Descriptive

B) Diagnostic

C) Predictive

D) Prescriptive

✓ Correct! Forecasting future values is Predictive analytics.

✗ "Forecast" indicates predicting future outcomes — that's Predictive analytics.

Quiz: Identify the Analytics Level

Q18: "The sales increase was driven by our new marketing campaign launched in October"

A) Descriptive

B) Diagnostic

C) Predictive

D) Prescriptive

✓ Correct! Explaining WHY something happened is Diagnostic.

✗ This explains the cause of an outcome — that's Diagnostic analytics (answering "why?").

Quiz: Identify the Analytics Level

Q19: "We recommend increasing marketing spend by 20% in Q1 to achieve the $2.5M target"

A) Descriptive

B) Diagnostic

C) Predictive

D) Prescriptive

✓ Correct! Recommending actions is Prescriptive analytics.

✗ "We recommend" indicates telling what to do — that's Prescriptive analytics.

3

The MAP Framework

Measure → Approach → Prerequisites

Selecting Your Analytics Technique: The MAP Method

A systematic approach for choosing the right analytical technique for Assessment 3:

M

Measure

"What am I trying to predict or explain?"

A

Approach

"Classification, regression, clustering, or association?"

P

Prerequisites

"Do I have what this technique needs?"

M — Measure: What's Your Outcome Variable?

The outcome variable determines which category of technique you need:

What is your outcome variable?

↓

Categorical (Yes/No, Groups)

Classification

Numerical (Continuous)

Regression

No outcome (exploratory)

Clustering / Association

            Key Question: Look at your business question from Assessment 1. What are you trying to predict? Is it a category or a number?
        

A — Approach: Technique Categories

Approach	When to Use	Techniques	Example Question
Classification	Predict a category	Logistic Regression, Decision Tree, Random Forest	Will customer churn? (Yes/No)
Regression	Predict a number	Linear Regression, Multiple Regression	What will Q1 sales be?
Clustering	Find natural groups	K-Means, Hierarchical	What customer segments exist?
Association	Find relationships	Association Rules, Market Basket	What products are bought together?

Quiz: Which Technique Category?

Q20: "I want to predict whether an employee will leave the company (Yes/No) based on satisfaction and tenure."

A) Classification

B) Regression

C) Clustering

D) Association

✓ Correct! Predicting Yes/No (categorical outcome) requires Classification.

✗ The outcome is categorical (Yes/No), so you need Classification.

Quiz: Which Technique Category?

Q21: "I want to forecast next quarter's revenue based on marketing spend and seasonality."

A) Classification

B) Regression

C) Clustering

D) Association

✓ Correct! Predicting a continuous number (revenue) requires Regression.

✗ Revenue is a continuous numerical outcome, so you need Regression.

Quiz: Which Technique Category?

Q22: "I want to identify distinct customer segments based on purchasing behavior."

A) Classification

B) Regression

C) Clustering

D) Association

✓ Correct! Finding natural groups without a predefined outcome is Clustering.

✗ You want to discover groups (no predefined outcome), so use Clustering.

Quiz: Which Technique Category?

Q23: "I want to discover which products are frequently purchased together to improve cross-selling."

A) Classification

B) Regression

C) Clustering

D) Association

✓ Correct! Finding items that appear together is Association (Market Basket Analysis).

✗ "Products purchased together" is classic Association Rules / Market Basket Analysis.

P — Prerequisites: Data Requirements

Each technique has specific data requirements. Check before committing:

Technique	Minimum Data Requirements	Data Quality Needs
Classification	Labeled outcome, sufficient examples per class	Balanced classes (or handle imbalance)
Regression	Numerical outcome, predictor variables	Check linearity, no extreme multicollinearity
Clustering	Numerical variables for distance calculation	Scaled/normalized data, no missing values
Association	Transactional data (items per transaction)	Sufficient transaction volume

Gap Analysis: If your data doesn't meet prerequisites, identify what's missing now — you may need additional data for Assessment 3.

MAP Framework Summary

M

Measure

Categorical → Classification
Numerical → Regression
None → Clustering/Association

A

Approach

Select specific technique within category

P

Prerequisites

Verify data meets requirements

Assessment 3 Connection: Use MAP to plan your methodology section. Be explicit about why you chose your technique and confirm your data meets the prerequisites.

Bringing It All Together

The complete journey from data to argument:

Phase	Framework	Output
Week 2-3	5W1H Method	Business question
Week 4	SPAR Method	Data insights & findings
Week 5	VINE Framework	Effective visualizations
Week 5	MAP Framework	Methodology plan for A3

            Your Path: Business Problem (5W1H) → Data Exploration (SPAR) → Visualization (VINE) → Methodology (MAP) → Assessment 3
        

Key Takeaways

VINE Framework

Variables — Know your data types first
Intent — Define what story you're telling
Narrative — Use the decision matrix
Evidence — Apply the three tests

MAP Framework

Measure — Identify outcome variable type
Approach — Select technique category
Prerequisites — Verify data requirements

For Assessment 3: Your descriptive findings become the foundation for predictive analysis. Use VINE to communicate findings and MAP to plan methodology.

Questions?

VINE for visualizations • MAP for methodology

Next: Apply these frameworks to your Assessment 3 planning