DS 102 Apply

What the final artifact is for

The Data Quality & Cleaning Notebook + Data Quality Report is DS 102's final artifact. You choose a real wrangling problem you actually face, and write the problem, audience, data, method, verification, output, limitations, governance considerations, human reviewer and next step. It becomes the starting point for later practice discussions.

What to fill in

• The problem, audience and data (source, grain, time coverage, key definitions)
• The method and workflow
• At least three verification checks
• Evidence and output
• Limitations
• Risk and governance considerations, as relevant
• The human reviewer
• The next step

A structurally complete Data Quality & Cleaning Notebook

A structurally complete artifact includes: a clear, checkable question; a stated data source and grain; a reproducible method; at least three verification checks; output that matches the claims; at least one limitation.

The five Practical Assessment tasks

The Data Preparation & Quality Practical Assessment has five tasks, in order: Inspect, Clean, Reshape, Join, Validate. Write your answers to all five directly into the Data Quality & Cleaning Notebook + Data Quality Report form below.

1. 1Task 1: Inspect
   State the schema, row counts, grain (what one row represents), a candidate key, a missingness summary, and any unit or definition risk you can see before changing anything.
2. 2Task 2: Clean
   Handle missing values, duplicates, invalid categories, wrong data types, and whitespace/parsing problems. For every change, answer: what changed, why, what evidence supports it, and could it destroy information.
3. 3Task 3: Reshape
   State the before grain, the reshape operation you applied, the after grain, and a uniqueness check proving the after-grain key is genuinely unique.
4. 4Task 4: Join
   State the expected cardinality, the pre-join and post-join row counts, unmatched keys, and duplicated keys, then diagnose any row multiplication using the worked example below.
5. 5Task 5: Validate
   Produce a Data Quality Report covering every field listed below, ending in an explicit pass/fail validation result.

• Row count
• Unique-key checks
• Duplicate count
• Missingness
• Unmatched keys
• Allowed values
• Ranges
• Units
• Date coverage
• Final grain
• Known limitations
• Validation result (pass or fail, stated explicitly)

How it is scored

A human reviewer scores your submitted artifact against seven weighted domains. The overall result is PASS at 80 or above, REVISION_REQUIRED at 65-79, and FAIL below 65 or whenever any critical-fail rule is triggered -- regardless of the weighted score. A PASS additionally requires the Join Cardinality domain and the Validation & Data Contract domain to each reach at least 2 out of 3.

• Many-to-many or unintended row multiplication is not detected (including the worked 15-to-45 example above)
• A unit mismatch materially affects the output and is not detected
• Destructive cleaning is performed without evidence or a retained trace
• The candidate-key uniqueness assumption is false and was never checked
• The Data Quality Report has no validation section at all

What you need to complete

• Complete both DS 102 modules and pass the post-test at 80% or above
• Submit the Data Quality & Cleaning Notebook + Data Quality Report (the Practical Assessment artifact) covering all five tasks
• Have that artifact scored against the Data Preparation & Quality Practical Assessment rubric with an overall PASS
• Trigger none of the five critical-fail rules above

What you will be assessed on

• The seven weighted rubric domains above: Inspection & Grain, Cleaning Decisions, Reshape Correctness, Join Cardinality, Validation & Data Contract, Lineage & Reproducibility, and Human Review & Limitations
• Join Cardinality must independently reach at least 2 out of 3 -- correctly diagnosing row multiplication is mandatory, not optional
• Validation & Data Contract must also independently reach at least 2 out of 3

Evidence produced

• Your submitted Data Quality & Cleaning Notebook + Data Quality Report
• A recorded rubric result (PASS, REVISION_REQUIRED, or FAIL) with no critical-fail rule triggered
• DS 101 completion is recommended background reading, but is never required to earn this certificate

What you can earn

The Data Preparation & Quality Capability Certificate, an L2 capability certificate with its own public verification page. It supplements your DS 102 Course Completion evidence -- it never replaces it, and it is never issued from the post-test alone.

What this credential does not mean

This credential confirms that the holder completed DS 102's required lessons and post-test, and demonstrated schema/grain, cleaning, reshape, join-cardinality, validation, and lineage capabilities against the Data Preparation & Quality Practical Assessment rubric.

• Does not confer a professional data-engineering license or employment qualification
• Does not certify that any specific dataset the holder cleans afterward is correct
• Does not remove the need for human review before high-impact decisions (Human, AI, Human)

Exit Ticket

Finally, use the Exit Ticket to record the most important idea from today, the step you are still unsure about, where you would apply this in a real context, the risk you most need to verify, and your difficulty and usefulness ratings.

Before you submit

Open the forms

• Fill in the Data Quality & Cleaning Notebook + Data Quality Report
• Fill in the Exit Ticket
• My Learning: Your progress