<aside>
⚠️ Teacher-only. Do NOT share this with students. It contains the exact scoring breakdown and reference values. A student who sees it can paste it into an LLM and reverse-engineer a passing submission without learning the material. The student-facing chapter carries only a softer ## How you will be evaluated section.
A copy is published as a standalone, unlisted Notion page for easy sharing with reviewers: Week 11 Assignment Rubric (TEACHERS). It lives outside the student-shared Data Track tree (under the platform root, not under any week) and is not linked from any student page. Set its share to "anyone with the link can view" and share the link only in teacher channels, never in a student-facing space.
</aside>
Use this rubric when reviewing student submissions for Assignment: Build Two Dashboards. One mark per criterion. Provide written feedback for any criterion scoring 0.
<aside> 📝 Two instruments, two scales. The repo's autograder is an automated CI check scored out of 100 (pass threshold 60): it verifies the mechanical, machine-checkable parts (files present, secrets hygiene, Streamlit code patterns, metric-definition coverage). It cannot verify the dashboards render real data or judge presentation quality. This teacher rubric is the holistic review scored out of 11: it adds the judgement calls a script cannot make (chart-type fit, metric-definition quality, presentation clarity). Use the autograder as a first-pass gate, then apply this rubric for the final mark. The two numbers are not meant to match.
</aside>
| Criteria | Points available |
|---|---|
| Metabase dashboard (3 Questions, correct SQL, chart types) | 3 |
| Metric definitions (five fields, all Metabase panels covered) | 2 |
Streamlit KPI panel (Postgres query, st.metric, cache) |
2 |
| Credentials in env vars, not in committed code | 1 |
| Streamlit freshness panel (row count + last updated from Postgres) | 1 |
| Presentation: what changed + why it matters + action, ≤5 min | 2 |
| Total | 11 |
Pass threshold: 6/11. Extra credit (Metabase date filter) is bonus : not scored, note qualitatively in feedback.
Award 1 point per Question that satisfies all of:
dev_<student>.fct_trips or dev_<student>.fct_daily_borough_stats with the correct schema prefix.0 points if: the Question uses public.fct_trips (missing schema prefix), returns all-NULL values for key columns, or shows placeholder hardcoded data instead of live query results.
Common issue: pickup_borough IS NOT NULL filter missing : half the rows may be NULL if the stg_zones join did not complete. Check whether dbt ran successfully before docking points here.