Guidelines, Timeline & Presentation
Department of Computer Science
University of the Philippines Cebu
One Question. One Dashboard. One Story.
"A dashboard that doesn't answer a question is just a pile of charts. A dashboard that answers one question well is a decision."
-- Your driving principle
An end-to-end analytics project that integrates everything from Weeks 1–11. You will frame a problem, acquire data, analyze it, and communicate findings via an interactive dashboard and a 12-minute presentation.
Every capstone revolves around a single driving analytics question.
Interactive, question-first, answers the question on first view.
12 minutes to walk the audience from question to recommendation.
Philippine context projects are encouraged but not required.
4 stages submitted through the Group Portal.
Solo work is allowed but not advised. 2 members is the recommended size. Team size determines expected scope:
You'll do everything alone. Allowed if you really want to, but pair up if you can.
One analytical method + interactive dashboard answering the driving question.
2+ analytical methods (e.g., regression + clustering) OR a multi-model comparison. Each member leads a distinct analytical component, documented in Stage 1.
| Aspect | Solo | Team of 2 | Team of 3 |
|---|---|---|---|
| Methods | 1 method | 1 method | 2+ methods OR comparison |
| Component leads | You do all | Shared | One per member (named) |
| Dashboard | Required | Required | Required |
| Presentation time | 12 min | 12 min | 12 min |
| Grading | Same rubric | Same rubric | Judged on proportional depth |
Students pick their own capstone topic. A good topic can be framed as one driving question, uses publicly available data, and is feasible within the remaining weeks.
| Weak topic | Strong topic |
|---|---|
| "Machine learning on climate data." | "Can rainfall in Cebu province predict dengue incidence 2 weeks ahead?" |
| "Cryptocurrency prediction." | "Do GCash transaction spikes on payday correlate with specific merchant categories?" |
| "Sentiment analysis of tweets." | "Which topics drive the most negative sentiment in PH political Twitter during elections?" |
PSA OpenSTAT • BSP • PAGASA • DOH • COMELEC • SEC EDGE • Kaggle (ASEAN)
Specific • Measurable • Actionable • Anchored in data you have
Your capstone revolves around one driving analytics question. This question must:
Can a stranger read your dashboard title, look at the first view, and understand what you found in under 60 seconds? If not, refine the question or the answer view.
Each stage is submitted via the Group Portal. Submissions include a presentation link, markdown summary, and optional PDF attachment.
Week 8–9. Define driving question, identify dataset, submit proposal. Teams of 3: name the 2+ methods and component leads.
Week 9–10. Exploratory data analysis, visualizations, handle missing values, engineer features, preliminary insights.
Week 10–11. Build + evaluate model(s). Document methodology, metrics, model choice. Teams of 3: evidence of 2+ methods.
Week 12. Ship interactive dashboard, deliver 12-min presentation + 3 min Q&A.
Each stage uses 5 guide questions in the portal to scaffold your submission.
Deadlines set by instructor per semester.
All team members present. Choose your format:
Traditional deck with dashboard demo in the middle.
Present directly from your dashboard. See next slide for the rules.
1–2 intro slides, then switch to dashboard.
| Segment | Time |
|---|---|
| Driving question & why it matters | 2 min |
| Data & methodology | 3 min |
| Results & dashboard demo | 5 min |
| Insights & recommendations | 2 min |
| Q&A | 3 min |
Rehearse with a timer. Over 12 minutes → points deducted on Communication.
Skipping slides? Your dashboard must carry the whole story. These rules are strictly graded:
Streamlit • Plotly Dash • Tableau Public • Power BI • Observable • Shiny
This mirrors how analytics is pitched in real business settings.
Each criterion is scored 1–4. Your final is the weighted sum.
| Criterion | Weight | Excellent (4) | Good (3) | Satisfactory (2) | Needs Work (1) |
|---|---|---|---|---|---|
| Problem / Question | 15% | Clear, relevant, well-scoped | Clear but could scope better | Somewhat unclear | Poorly defined |
| Methodology | 25% | Well-executed; 3p teams: 2+ methods | Appropriate, minor issues | Some issues | Inappropriate / missing |
| Analysis Quality | 30% | Deep insights, rigorous | Good insights | Surface-level | Missing insights |
| Visualization / Dashboard | 15% | Professional, question-first | Good, minor issues | Basic | Poor / missing |
| Communication | 15% | Engaging, clear, well-timed | Clear, low engagement | Some confusion | Hard to follow |
3-person teams are graded on proportional depth — expect more scrutiny on Methodology and Analysis Quality.
No bonus tracks — just the rubric.
Every stage submission on the Group Portal needs the same basics. Run through this before hitting Submit.
Yes, but it's not advised — you'll do everything alone. Pair up if you can.
Yes, but by Stage 2 at the latest. After that you risk losing points on Problem definition.
No. Philippine context is encouraged but not required.
Prepare backup screenshots or a fallback deck. No time extensions for tech failures.
No. Each member must lead a distinct analytical component, documented in Stage 1, evidenced in Stage 3.
Yes, but justify the choice and explain them clearly in your presentation.
Full guidelines: docs/CMSC178DA_Capstone_Guidelines.md
The capstone isn't about proving you can run sklearn. It's about proving you can take a messy question, frame it precisely, and deliver an answer that someone could act on.