Teacher guide — Grade 10, Unit 7: Dashboards with Streamlit¶

Time: 3 lessons + project (about 5 periods). Track: B (code).

#	Lesson	Time
1	Design on paper	40–45 min
2	Build with Streamlit	40–45 min
3	Publish and review	40–45 min
P	Unit project: Mini dashboard	2 periods

Before you start¶

Materials: app/air_dashboard.py and requirements-streamlit.txt in this unit's folder. Install: pip install streamlit (internet needed once). Run: streamlit run app/air_dashboard.py.
Read docs/pedagogy/lesson-anatomy.md and docs/pedagogy/ai-use-policy.md.

If you have no computers / no internet¶

Sketch a dashboard on paper: title, one chart, one selector, source note. Peer-test with 'tell me what you see'.

Common misconceptions¶

More charts = better dashboard. One clear message beats ten charts.
Dashboards can't be wrong because they're interactive.

Differentiation¶

Support: give a partially filled table / sentence starters; pair with a data-steward role.
Stretch: ask "what would change your conclusion?" and "who is missing?"

Answer key (check questions)¶

Lesson 1: Design on paper 1. What is the first design decision?
Answer: The question the dashboard answers. 2. Why add a source note?
Answer: So viewers can judge trust. 3. How many main messages?
Answer: One.

Lesson 2: Build with Streamlit 1. How do you run an app?
Answer: streamlit run file.py 2. What does st.sidebar.selectbox do?
Answer: Adds a dropdown to the sidebar. 3. What does the script do when you click?
Answer: Reruns from top with the new value.

Lesson 3: Publish and review 1. Name two things to check before sharing.
Answer: Source and limits; no personal data; licence. 2. Does hosting change the data's licence rules?
Answer: No: attribute and respect share-alike. 3. Why test on a phone?
Answer: Many users have only phones.

Assessment¶

Starter quiz each lesson (3 questions, one recalling the previous lesson); unit quiz in quiz.yml; project marked with the rubric in docs/pedagogy/assessment.md; 2-minute oral check per group.

Alignment¶

Optional Computer Science Grade 10: Python programming, databases and AI & contemporary technology; compulsory Maths data handling/statistics.

Answers: practice exercises¶

User: principal; question: "Which classes had the lowest attendance this month?"; decision: whom to follow up; 10-second view: school attendance % this month and a bar chart by class.
st.metric shows one big number (with optional change); st.line_chart draws a chart from a table/series.
So the code can be written top-to-bottom like a normal script; each run uses the widgets' current values — no complicated event handling.
Title = question; metric: belt or national rate; main chart: horizontal bars of the top 10 districts by rate; widget: belt or province selector; note: synthetic data.
belt = st.selectbox("Ecological belt", sorted(df["ecological_belt"].unique())) and st.metric("Districts in belt", int((df["ecological_belt"] == belt).sum())).
st.metric("Mean PM2.5, 2020 (µg/m³)", "52.8") (plus "synthetic" in the caption).
A long default smoothing window that hides peaks; auto-scaled y-axes that exaggerate small changes; a flattering default period or place.
Do not offer downloads of personal/sensitive data or data whose licence forbids redistribution (e.g. DHS); credit and licence must travel with any download.
Show only aggregated counts/rates by belt or district with a small-group rule; never individual records, ages with locations, or anything identifying.
Poster: works offline, one clear message, easy to print; dashboard: users explore their own question, updates with new data. (Poster: fixed; dashboard: needs a device and internet.)

Real-data exercises: notes¶

R1 Check that share-alike is acknowledged and that the app does not re-host the full raw dataset publicly unless the maintainers agree (D7/D8).
R2–R3 Encourage saving downloaded CSVs locally and loading them with @st.cache_data; don't call APIs on every rerun (Open-Meteo free tier limits).