Data Analyst Portfolio Project

Convert a stakeholder question into scope

Convert a stakeholder question into scope

Use the existing Foundations or Applied capstone as a prerequisite and choose a new question or extension. For Haven, ask how completed value, order count and AOV differ by period or region. State the unit, population, date coverage and missing business definitions before coding. Do not claim recognised revenue or causal impact from line values. A project plan should name the evidence needed for a decision as well as the available data, so the conclusion can honestly be “more evidence required.”

Follow a concrete example

Identify the reader’s decision before selecting a chart. The completed-value contract measures supplied historical line amounts, not net accounting revenue. Region comparisons need a compatible distinct-order denominator. Request missing business evidence explicitly rather than substituting an attractive metric that happens to be present in the file.

Further reading: DataCareerHub learning quality. The case and exercises are authored for DataCareerHub.

Your task

Draft the brief, metric contract and evidence request. Choose a region comparison as an independent extension.

Use learner_worksheet.txt and the provided source/code files. Save your own artifact and compare the separate solution after attempting the task.

Integrate SQL/Python analysis with reporting

Integrate SQL/Python analysis with reporting

Choose SQL or Python according to the method you can reproduce and explain. Keep the same population contract across the query, chart and prose. Validate joins, count distinct orders and reconcile independent aggregates. Use the existing BI/dashboard course for interaction mechanics instead of duplicating it. Show the effect of a controlled source mutation and preserve its status as a scenario. Your own implementation and extension should remain editable and distinguishable from supplied reference artifacts.

Follow a concrete example

Build an evidence matrix with baseline, join-grain, changed-input and unavailable-period cases. The42-row joined result and21-order count should coexist without relabelling one as the other. A valid quantity change affects value but not the number of order keys. Keep the edited source in a separate scenario so baseline evidence remains traceable.

Inspect the reference implementation

Language: Python 3.12. The complete downloadable file includes the named helpers and the fresh-output command-line runner. Trace the source, transformation and result against this lesson’s contract before running it locally.

def career_evidence(source,out):
 base=py_project(source,out)
 return {'project_claim':'Implemented an analysis of fictional Haven Workshop orders with reconciled line and order grain.','evidence':base,'avoid_claims':['client delivery','employment outcome','accreditation','causal sales improvement'],'independent_extension':'Compare region-level order counts and value, retaining cancelled-order exclusions.'}

Further reading: DataCareerHub learning quality. The case and exercises are authored for DataCareerHub.

Your task

Build the analysis and one visual/table. Include row/order reconciliation, a mutated-input test and an empty-coverage state.

Use learner_worksheet.txt and the provided source/code files. Save your own artifact and compare the separate solution after attempting the task.

Write a decision note and defend the limits

Write a decision note and defend the limits

A decision note separates finding, context, limitations and next action. April value is EUR 600 versus May EUR 590, while counts and AOV move differently. Explain the trade-off and missing evidence before proposing an action. Avoid overstating a small two-month snapshot. Include a reviewer challenge and how your validation answers it. A portfolio deliverable can show careful restraint and an explicit next evidence request; it need not manufacture a dramatic improvement. No instructor review or certification is implied unless actually provided.

Follow a concrete example

A useful decision note reports what changed, what else changed and what is still unknown. Higher AOV alongside lower value/count is a prompt to investigate mix, not a contradiction to hide. Ask a reviewer to challenge one denominator or exclusion and show the artifact answering that challenge. The suggested self-review score has no accreditation status.

Further reading: DataCareerHub learning quality. The case and exercises are authored for DataCareerHub.

Your task

Write the final note and score scope, method, checks, communication and reproducibility 0–2 each with linked evidence.

Use learner_worksheet.txt and the provided source/code files. Save your own artifact and compare the separate solution after attempting the task.