Data Analytics Capstone Project: Complete a Case Study
Core concepts
The capstone asks you to run the full lifecycle on a self-chosen dataset and present it as a portfolio case study. The work is judged on process as much as outcome.
Ask - state the business task in one sentence, identify stakeholders, and note assumptions and constraints. Prepare - describe the data source, apply ROCCC, document licensing or privacy limits, and record the row and column counts. Process - list every cleaning step: duplicates removed, nulls handled, types corrected, values recoded. Keep an untouched copy of the raw data and a cleaned copy. Analyze - use at least one aggregation and one comparison, state the metric definition, and cross-check totals. Share - build a small set of visualizations, each tied to a finding, with a headline that states the takeaway rather than the chart type. Act - recommend a next step the stakeholder could actually take.
Deliverables typically include a written summary, the notebook or spreadsheet with reproducible steps, and a slide deck or dashboard. Hosting code on GitHub or Kaggle and writing a short README with the question, data source, method, and findings makes the work reviewable.
Traps
Common failures: skipping the Ask phase and analyzing whatever the file contains, leaving cleaning undocumented, presenting a chart with no stated takeaway, or claiming causation from a correlation. Reviewers also penalize conclusions that exceed the data - for example, generalizing from a convenience sample.
Rule of thumb
If a stranger cannot reproduce your result from your notes, the case study is incomplete. Write each cleaning step as if explaining it to a new teammate.