Learning Objectives

By the end of this lesson, learners should be able to:

  • Conduct exploratory data analysis (EDA).
  • Summarize datasets using tables and charts.
  • Identify trends, patterns, and anomalies.
  • Translate statistical findings into business recommendations.

What Is Exploratory Data Analysis?

EDA is the process of investigating data before formal modeling. It helps analysts understand structure, quality, relationships, and unusual observations.

John Tukey described EDA as an approach for “letting the data speak.”

Typical EDA Workflow

  1. Inspect data types.
  2. Check missing values.
  3. Generate summary statistics.
  4. Create charts.
  5. Identify outliers.
  6. Explore relationships.
  7. Document findings.

Example Dataset

A global hotel chain analyzes occupancy rates for hotels in London, Dubai, Singapore, Toronto, and Sydney.

Summary Statistics

  • Mean occupancy = 78%
  • Median occupancy = 80%
  • Standard deviation = 6%
  • Minimum = 52%
  • Maximum = 89%

Interpretation

One hotel with occupancy of 52% is substantially below the portfolio average and should be investigated.

Business Interpretation

Possible explanations for the low-performing hotel include:

  • Increased local competition,
  • Pricing problems,
  • Service quality issues,
  • Renovation disruption,
  • Economic downturn in the local market.

Recommended Actions

  • Review pricing strategy,
  • Increase marketing activity,
  • Analyze customer reviews,
  • Assess operational performance.

Relationship Analysis

Scatter plots can reveal relationships between variables.

Example

Marketing expenditure vs. room bookings across hotels.

If higher marketing spending is associated with higher bookings, management may consider increasing promotional investment.

Communicating Results To Management

Effective communication should include:

  1. Key statistic,
  2. Business meaning,
  3. Possible cause,
  4. Recommended action.

Example

“Average occupancy is 78%, but the Toronto property is significantly below the portfolio average at 52%. Management should investigate local market conditions and pricing strategy.”

EDA Visualization Toolkit

Purpose

Recommended Chart

Trend over time

Line chart

Category comparison

Bar chart

Distribution

Histogram

Outlier detection

Box plot

Relationship

Scatter plot

International Case Study: E-Commerce Company

A global e-commerce company analyzes customer order values.

Findings

  • Mean order value: USD 95
  • Median order value: USD 62
  • Strong positive skewness,
  • Top 5% of customers contribute 38% of revenue.

Business Actions

  • Develop premium loyalty program,
  • Personalize marketing for high-value customers,
  • Protect key accounts from competitor targeting.

Common EDA Mistakes

  • Ignoring missing values.
  • Focusing only on averages.
  • Overlooking seasonality.
  • Failing to validate unusual results.
  • Confusing correlation with causation.

Practical Activity

Using a sales dataset:

  1. Calculate descriptive statistics.
  2. Create a histogram and box plot.
  3. Identify outliers.
  4. Analyze one relationship using a scatter plot.
  5. Prepare a one-page management summary.

Learning Materials / Reference Materials

  • Tukey, J. W. Exploratory Data Analysis.
  • Pandas and Excel EDA tutorials.
  • Tableau and Power BI EDA resources.

Lesson Summary

EDA helps analysts discover patterns, quality issues, and business insights before advanced modeling begins. Effective interpretation converts statistics into actionable managerial recommendations.

PreviousNext