Track 14 · Practical Data Analysis · Lesson 8 of 11
An exploration workflow
Explore a dataset in a disciplined order — question, data, one-variable summaries, charts, relationships — and turn what you see into tentative hypotheses rather than conclusions.
A manager forwards a salary file and asks: do women earn less here? You could open a chart, spot a gap and reply in five minutes. Following a short workflow — and checking one more variable before you answer — is what separates a finding you can defend from one that falls apart in the first meeting.
Exploratory data analysis (EDA) is the open-minded first look at a dataset: summaries and charts that show what is there, what is odd, and what might be worth testing. It is not a free-for-all. Without an order, people jump to the most eye-catching chart, stop as soon as something looks interesting, and report it as a result.
The workflow below keeps you honest. Each step answers one question before the next begins.
Included with a Formula Quest membership
It has 8 graded exercises with feedback, hints and worked solutions. After it you will be able to:
- Follow a fixed order: question, data, one variable, groups and relationships, third variables, hypotheses
- Check whether a third variable (such as job level) explains a difference between groups
- Explain why many unplanned comparisons produce false alarms, and estimate how often
- Write findings from exploration as tentative hypotheses with a plan to confirm them