Day 04 Part 2 — Exploratory Data Analysis: Agenda¶
EDA is how you go from "I have clean data" to "I understand what's in it." Before you build dashboards or write reports, you need to explore — look for patterns, anomalies, relationships, and surprises that raw numbers hide. EDA is detective work: you're looking for the story the data is trying to tell.
Session Overview¶
Duration: 3 hours Prerequisite: Data Cleaning (clean data is the starting point for EDA) Tools: Pandas, matplotlib, seaborn in Jupyter Lab
Learning Objectives¶
By the end of this session you will be able to:
- Compute and interpret descriptive statistics
- Perform univariate analysis: understand the distribution of individual variables
- Perform bivariate analysis: understand relationships between pairs of variables
- Identify and visualise trends over time
- Compute and interpret correlation
- Structure an EDA workflow and communicate findings
Session Flow¶
| Time | Topic | File |
|---|---|---|
| 0:00 – 0:30 | Descriptive statistics | 03-descriptive-statistics |
| 0:30 – 1:00 | Univariate analysis — distributions | 04-univariate-analysis |
| 1:00 – 1:30 | Bivariate analysis — relationships | 05-bivariate-analysis |
| 1:30 – 2:00 | Trend analysis — time series EDA | 01-trend-analysis |
| 2:00 – 2:30 | Correlation analysis | 02-correlation-analysis |
| 2:30 – 3:00 | EDA case study — end to end | 06-eda-case-study |
EDA Mindset¶
EDA is not report writing. It's exploration. The goal is to: - Understand the data structure — what columns exist, what they mean, how they relate - Find the distribution — where are values concentrated, what's rare, what's impossible? - Spot anomalies — anything that doesn't fit the pattern is worth investigating - Identify relationships — which variables move together? - Generate hypotheses — "I notice X correlates with Y. Is that causal?"
Analyst wisdom
"Torture the data, and it will confess to anything." — Ronald Coase
EDA is how you avoid that trap. Look at the data before deciding what question to answer.
← Data Cleaning Interview Questions · Next: Descriptive Statistics →