Capstone 3992x300f.png

⭐️

Assignment Overview

Data Summary

The data summary demonstrates that all data necessary to answer your research question has been gathered and organized, or that pipelines are implemented that gather and organize data without further manual intervention. After this milestone, you should not need active work to bring in new data. Analysis, modeling, and communication may continue, but ingestion should be stable.

Authoritative due date: the Canvas assignment (typically week 7 in the published schedule).

📋

Required Contents

Submit a PDF or instructor-approved format that includes all sections below. Point to repository paths, scripts, and README sections wherever possible.

  1. Research question reminder. One paragraph linking the frozen data scope to the question from your proposal.
  2. Data inventory. Table of datasets or tables: source, license/terms, row counts or size, date range, refresh policy, and location in the repo or warehouse.
  3. Schema and organization. Logical schema, file layout, or ERD; primary keys and join paths; how raw vs analysis-ready layers relate.
  4. Ingestion status. Manual pulls completed, or automated pipeline described (schedule, orchestration, failure handling). Evidence that runs succeed without babysitting.
  5. Data quality and profiling. Missingness, duplicates, outliers, or label issues you found; steps taken or documented caveats for analysis.
  6. Reproducibility. Commands or Makefile targets to rebuild the analysis-ready dataset from documented inputs; environment notes (Python/R versions, secrets handling).
  7. Ethics and access controls. PII handling, redaction, consent, or sharing constraints still in force.
  8. Freeze statement. Explicit confirmation that no further ingestion work is planned; if the proposal changed, summarize deltas and instructor approval.

Submission Checklist

📐

Grading Rubric

Scored out of 100 points.

Criterion Description Points
Completeness Vs Research Question Data needed to answer the question are present, documented, and scoped correctly. 25
Organization And Schema Clear structure, keys, and layers; a peer could navigate the data without guesswork. 20
Pipelines Or Stable Ingestion Automated or one-shot ingestion is reliable; no ongoing manual firefighting required. 20
Quality And Profiling Honest assessment of data flaws; mitigations or documented limitations for analysis. 15
Reproducibility Documented, tested path from source to analysis-ready data. 10
Ethics And Freeze Responsible-use notes are accurate; freeze statement is credible and aligned with repo state. 10
Total 100