Skip to content

MODULE 5 · LESSON 28 · Analytical thinking

Identifying patterns

Distinguish a repeatable descriptive pattern from a coincidence or an artifact of selection.

By Laurence Svekis · Self-paced · All worked data is synthetic unless you supply your own source.

Try the practice ↓

PredictInspectCalculateCheckExplainTransfer

Your starting point

Prerequisites: Use the ideas and vocabulary from lesson 27; revisit its worked examples if needed.

Definition: A pattern is a regularity in observed data, such as a trend, cycle, cluster or association. Pattern detection is exploratory unless a specific expectation and evaluation method were established beforehand. Repetition in a small sample may occur by chance or by construction.

Look for structure with context

Plot values in an order that makes sense, inspect categories and compare summaries. Ask whether the pattern could result from data collection: weekday-only records, repeated users or a changed reporting definition. A numerical regularity may describe the measurement process rather than the phenomenon of interest. Synthetic teaching data can contain deliberately constructed patterns and should never be treated as empirical discovery.

Beware many opportunities to find something

If you inspect enough groups, periods and metrics, some striking contrast is likely to appear even without a stable underlying effect. Keep a record of the comparisons you tried. Treat a surprising exploratory result as a hypothesis to examine with new evidence rather than presenting it as if it were the only question considered.

Test whether the pattern survives a reasonable change

Compare adjacent periods, relevant subgroups or a held-out portion where feasible. Check whether one unusual record creates the whole story. A stable pattern across these checks is more informative, but it still does not automatically establish causation or future reliability. Describe the observation precisely and identify the boundary of the data that supports it.

Two worked examples

A calendar artifact

A website table shows lower traffic every seventh record. If the records are daily and complete, a weekly cycle is plausible. If weekends were omitted or dates are irregular, record position is not the same as calendar time. Inspect actual dates before labeling the pattern weekly seasonality.

One influential order

Monthly revenue rises sharply because of one unusually large wholesale order. The total increase is real if the order is valid. Calling it a broad increase in ordinary customer demand would need additional evidence. Compare counts, typical order size and the contribution of that record.

Where this becomes useful

A project lead sees delays concentrated in review tasks. That may suggest a bottleneck, but task definitions, dependencies and reporting practices could also create the cluster. Use the pattern to direct investigation rather than to assign blame.

Make an attempt

Open the complete workbench with Project management. Compare actual hours by team and filter to completed tasks. Write a pattern you observe, then a collection or task-mix explanation that could produce it. Check whether the pattern depends on one row.

Open the accompanying interactive lab →

Misconception check

Claim: A visible pattern must have an important real-world explanation.

Inspect the correction

It can arise from chance, data construction, measurement rules or selection. Investigate before assigning meaning.

Independent application

Find a pattern in a synthetic dataset, document the comparisons you tried, and perform two robustness checks such as changing a reasonable filter or inspecting a relevant subgroup. Write a descriptive finding and a separate follow-up hypothesis.

Completion records practice and self-review. It is not certification of statistical competence.