Lesson 4 of 4

Visualisation, dashboards and presenting insights

2 min read

Choose the chart for the question

Question	Chart
How does it change over time?	line chart
How do categories compare?	bar chart (horizontal for long labels), sorted
What's the distribution?	histogram or box plot
How do two numbers relate?	scatter plot
What share of a whole?	stacked bar or a simple table; pie only for 2–3 slices
Where do people drop off?	funnel

Rules: start bar axes at zero; label axes and units; use one highlight colour for the point you're making and grey for the rest; remove gridlines and 3D effects; title charts with the takeaway ("Hire rate doubled after adding tasks"), not the metric name.

python

import matplotlib.pyplot as plt
monthly = df.groupby("month")["job_id"].count()
ax = monthly.plot(kind="line", marker="o", figsize=(8, 4))
ax.set_title("Applications grew 35% after the campus drive in August")
ax.set_ylabel("Applications"); ax.set_xlabel("")
plt.tight_layout(); plt.savefig("applications.png", dpi=150)

Dashboards that get used

One audience and one purpose per dashboard (leadership summary vs operations tracker).
Top row: 3–5 KPIs with the comparison that gives them meaning (vs target, vs last month).
Then trends, then breakdowns, then detail tables. Filters for date, region and product.
Consistent definitions documented in the dashboard ("Active = logged in within 30 days").
Fast: pre-aggregate heavy data; avoid 30 visuals on one page.

In Power BI know: data model and relationships (star schema: fact table + dimension tables), measures in DAX (Total Sales = SUM(Sales[Amount]), CALCULATE with filters), slicers, and scheduled refresh. In Tableau: dimensions vs measures, calculated fields, LOD expressions, dashboards and actions.

Presenting: answer first

Structure every presentation or email like this:

The answer: "Applications fell 20% because the mobile apply button broke on 3 March."
The evidence: the chart showing the drop on mobile only, starting that day.
The impact: "about 1,800 lost applications, mostly from tier-2 cities."
The recommendation: "fix and add an automated check on the apply flow; re-invite the affected students."

Know your numbers, say what you're unsure about, and keep the methodology in the appendix.

A portfolio that gets interviews

Build 2–3 projects on real public data (government open data, Kaggle, your own scraped data): a clear question, cleaning notebook, SQL queries, a dashboard and a one-page summary of findings. Put them on GitHub with a readable README, and link a live dashboard.

Check yourself

Which chart would you use to show the share of applications by city over 12 months?
Write a chart title that states a takeaway.
Structure a 2-minute update on why revenue dropped last week.

← Previous