Your starting point
Prerequisites: Use the ideas and vocabulary from lesson 58; revisit its worked examples if needed.
Definition: An analytical workflow is an ordered, documented process connecting a decision question to evidence and a defensible conclusion. Its stages depend on one another: a chart cannot repair incorrect source definitions, and a polished report cannot repair an invalid denominator.
Plan and preserve
State the question, prediction, population, period and evidence of success. Inspect provenance and the row grain before editing. Preserve the source and identify required fields, missingness and duplicate rules. Choose transformations for documented reasons rather than cleaning until the table looks convenient. A small analysis plan reduces the temptation to change the question after seeing a preferred result.
Analyze with checkpoints
After each transformation, reconcile counts and totals that should remain stable. Verify one row-level formula and one aggregate independently. Use a chart that matches the question and compare its plotted values with the summary table. Inspect relevant subgroups and unusual observations without turning every exploratory contrast into a confirmed finding. Keep the current filters visible in the report.
Communicate and test transfer
Write a scoped finding, a limitation and a proportionate next step. Export the cleaned table, settings and report, then restore them to check reproducibility. Finally, apply the same reasoning to a changed dataset without copying the original conclusion. Transfer means knowing which steps remain useful and which definitions or assumptions must be reconsidered for the new context.
Two worked examples
A messy invoice workflow
The synthetic operations table has 21 rows, including one exact repeated record and one missing hours value. After confirming the repeat and trimming whitespace, keeping one copy leaves 20 rows. Calculating hours times rate leaves one result unavailable. Report 19 valid billable calculations rather than inventing hours for the missing record.
A chart-to-report check
A region chart shows grouped revenue sums. Before writing the finding, reconcile the group sums with the filtered total and confirm that the chart did not include a subtotal row. The report then names the synthetic source, filter, unit and missing-data policy instead of presenting the chart as self-explanatory.
Where this becomes useful
A beginner can make a useful contribution by producing one small, traceable analysis that another person can inspect. Reliability comes from the chain of checks, not from using the largest dataset or most advanced chart.
Make an attempt
Run the complete workbench on Small business operations. Record a question, inspect quality, remove only the confirmed exact repeat, calculate a value, build a suitable chart and export a report. Keep the missing hours explicit and document unresolved category differences.
Misconception check
Claim: A workflow is complete when a chart appears.
Inspect the correction
A defensible workflow also verifies values, explains limitations, preserves the method and checks reproducibility.
Independent application
Complete a compact analysis on a second synthetic dataset using the same workflow. Document which definitions changed, verify one formula and one chart value, and explain the result without consulting the first project’s conclusion.
Completion records practice and self-review. It is not certification of statistical competence.
