← Courses

Free mini-course · 45 minutes · No account needed

Check an intake file without hiding rejected or corrected records

Apply a fixed data contract, quarantine conflicting records, then reconcile a documented correction. Build a release review that accounts for every input row.

For someone preparing small operational files. Read text IDs as text and use a calculator; no spreadsheet formulas or SQL are required.

What you will make

A three-part data release portfolio with accepted and held row IDs, reconciliation totals and a correction note that preserves the original evidence.

The exercise contract

All records, policies and outcomes in this course are fictional. The supplied contract determines the expected answers.

  1. All records and source documents are fictional. Keep raw rows and their row IDs; do not overwrite them while reviewing.
  2. For unit 1, account_code must contain exactly four ASCII digits, including leading zeros. quantity must be a whole number from 1 through 9.
  3. For unit 2, a repeated event_id holds every row sharing that ID. A row with an account absent from the supplied unique account list is also held. Only rows passing both checks may be released.
  4. For unit 3, the verified source-document quantity is the authorized correction input. Retain original and corrected quantities, document ID and revision; do not silently overwrite an issued file.

Classify a row before allowing it into the accepted total.

1. Apply the contract and keep a hold list

A data contract states what a field is allowed to contain. It does not tell you to make an invalid value look plausible. In this exercise an account code is four digits stored as text, so 0012 and 12 are different representations. Preserve the raw code, validate its shape and keep the row ID beside any decision.

Check presence, format and range separately. A missing account needs clarification; an explicit quantity of 0 fails this particular contract because the allowed range starts at 1. A decimal quantity fails the whole-number requirement. These outcomes come from the supplied contract, not a universal rule that zero or fractional quantities are always bad data.

Build accepted and held lists before summing quantities. A calculation over all rows can be numerically correct yet include records that are not ready for use. Report why each held row was excluded from the accepted total. Keep every input row represented in the review so a rejected record does not disappear from the handoff.

A hold is a work item, not a correction. Assign a follow-up to confirm the source value rather than filling a blank or rounding a number by guesswork. The final release can be smaller than the intake file while still accounting for all rows. This audit trail is the useful output of this unit.

Try it yourself

Classify the six rows under the unit-1 contract. Sum only quantities on accepted rows.

Raw intake rows

row idaccount codequantity
D100124
D203450
D33
D401232.5
D504562
D607897

Numbers use the units in each label. Blank answers are incomplete. ID lists can use commas, spaces or semicolons.

Write a handoff naming the accepted total and a source-confirmation action for each type of held row. Your explanation is saved but not automatically graded. Use fictional information only.

Complete the fields, then check your work.

Read this exercise’s solution

Accept D1, D5 and D6, total 13. Hold D2 for the zero quantity, D3 for the missing account and D4 for the fractional quantity. Three accepted plus three held equals all six input rows.

Apply the idea

If a new contract allowed quantity 0, which decision would change and which raw evidence should remain unchanged?

Avoid choosing a convenient duplicate or hiding an unmatched reference.

2. Quarantine conflicts before joining to a reference list

A syntactically valid row can still be unsafe to combine with other rows. The event ID in this unit is meant to identify one event. When the same ID appears with two different quantities, selecting the first record is a business decision disguised as a technical shortcut. The exercise therefore holds every row sharing a repeated event ID.

Check the supplied account reference separately. It is a unique list, and a missing match means that the account needs investigation. An inner join that simply drops unmatched rows would make the resulting report look clean while concealing the unresolved record. Keep the row ID and a clear unmatched-reference reason in the hold list.

Apply both checks to the original batch, then release only rows with a unique event ID and a known account. Keep row-level decisions separate from the accepted quantity total. If a row fails more than one rule, record all applicable reasons without counting it as two different input records.

Reconciliation is the final control: accepted rows plus held rows must cover the batch exactly once. A separate count of distinct event IDs can help explain duplicates, but it is not interchangeable with row count. Your note should make clear whether a number describes records, events or quantities.

Try it yourself

The reference list contains 0012, 0345 and 0789, each exactly once. Classify the batch using repeated event IDs and account membership.

Events awaiting reference checks

row idevent idaccount codequantity
J1E1000124
J2E1103452
J3E1103453
J4E1299991
J5E1300125
J6E1407892

Numbers use the units in each label. Blank answers are incomplete. ID lists can use commas, spaces or semicolons.

Explain why keeping the first E11 row or dropping 9999 in a join would hide an unresolved decision. Your explanation is saved but not automatically graded. Use fictional information only.

Complete the fields, then check your work.

Read this exercise’s solution

Hold J2/J3 for E11 and J4 for unknown account 9999. Release J1/J5/J6 with quantity 11. The row count reconciles: three released plus three held equals six.

Apply the idea

What evidence would be needed to authorize one E11 value before either row could move out of quarantine?

Preserve the old value and show why the new total changed.

3. Release a documented correction and reconcile its effect

A source check can turn a hold into an authorized correction. The verification table below records the original quantity, the checked source quantity and a document ID. Compare the values explicitly. Do not label an entire file verified merely because one disputed record was checked.

Calculate the original total and the source-verified total separately. The difference is the correction effect, not new operational activity. A negative change means the previous intake overstated quantity. Showing both totals helps the receiving team distinguish a revision from an additional delivery or transaction.

Keep a correction log containing row ID, old value, new value, document reference and version. If the previous output was already distributed, a new filename alone does not tell recipients which data supersedes which. Identify the affected row and tell the owner to confirm the replacement before downstream use.

This exercise provides the authority to use the verified source quantity only. It does not authorize inventing a missing document, merging unrelated accounts or changing an operational system. The portfolio should retain enough evidence for another reviewer to reproduce the revised total and see the release decision.

Try it yourself

All IDs and accounts in this new three-row fixture are valid and unique. The verified source quantity is authoritative for this exercise. Report corrected minus original for the change.

Document-verified correction evidence

row idevent idoriginal quantityverified quantitydocument id
R1F1044DOC-1
R2F1162DOC-2
R3F1211DOC-3

Numbers use the units in each label. Blank answers are incomplete. ID lists can use commas, spaces or semicolons.

Write the correction note with R2’s old/new values, its source document, the total effect and the owner who must confirm replacement. Your explanation is saved but not automatically graded. Use fictional information only.

Complete the fields, then check your work.

Read this exercise’s solution

R2 changes 6→2 using DOC-2. Original total 11; verified total 7; effect −4. Preserve the original draft and issue an identified corrected version with recipient confirmation.

Apply the idea

What extra action is needed if the receiving team has already imported the earlier draft?

Save your three-part portfolio

Check all three exercises and include an explanation for each. You may download a portfolio with mismatches for later review; it clearly records which checks passed. Editing an answer or explanation invalidates that exercise’s previous check.

Complete and check all three exercises to prepare your portfolio.

Inspect the portfolio before downloading
Your portfolio will appear here after each exercise has been checked.

Download before leaving. Course answers stay in this page and are cleared on refresh; they are not sent to a server or added to the older course progress tracker.

Source and scope

Occupational task context comes from the O*NET® 31.0 Database, USDOL/ETA, under CC BY 4.0. View the occupation.

Task 11402 · 08/2018 · Compile, sort, and verify the accuracy of data before it is entered.

Task 11403 · 08/2018 · Compare data with source documents, or re-enter data in verification format to detect errors.

Skill Workbench selected this context and authored the teaching, examples, practice rules, data and feedback. USDOL/ETA has not approved, endorsed or tested these additions. O*NET® is a trademark of USDOL/ETA. Matching the fixed checks does not certify occupational competence.