Historical Online Retail: define cohort, missingness and RFM policy before interpreting segments
Management & operations · Analytics & data · L3 · en
- Suitable for
- Readers planning a bounded historical retail analysis who meet the stated knowledge prerequisites.
- Learning goals
- State invoice-line grain,sterling units,historical period and timestamp timezone unknown.
- Define disjoint cohort exclusions and distinguish them from overlapping cancellation/negative/missing diagnostics.
- Define recency/frequency/positive-sales monetary using a cutoff and tie-preserving quartile policy.
- Explain why numeric CustomerID/history is not anonymity and why source no-missing metadata differs from CSV empty fields.
- Write a bounded historical analytical proposal with denominator/assumption/limitation statements, without observed effectiveness claims.
- Knowledge prerequisites
- CSV parsing and row/entity grain; Grouped aggregation and distinct-invoice counting; Dates, missing-field handling, exact monetary arithmetic and cancellation rules; RFM meanings and limitation of historically chosen policy
- Pathline suggested preparation
- Read the exact UCI dictionary and dataset-specific licence before choosing whether to analyse the original source download.; Specify the cohort, cutoff, duplicate policy and tie rule before aggregation. Python Decimal is one optional exact-arithmetic implementation.
- Duration
- 180 minutes (Pathline estimate)
- Source practice
Source exercises: Not confirmed
Source feedback: Not confirmed
Executable practice: Not confirmed
- Source answers
- Not applicable
This status applies only to this unit; check its limitations for task coverage. Feedback is listed separately.
- Access for this unit
- Public and free. The reviewed source page was readable without registration; use the original UCI source for dataset access. Local analysis tools are a separate choice.
- Environment
- Platform requirements not confirmed. Read the page in a browser. Optional analysis requires a tool capable of handling CSV, dates, grouping and exact monetary arithmetic; no Pathline code, data download or hosted runtime is supplied.
Next action: Read the UCI Variables Table, License and Citation sections. If analysing the data, follow the original UCI download link and first write an invoice-line grain, cohort-exclusion, cutoff, duplicate and tie policy.
Open this source unit ↗Pathline practice suggestions
These are suggested actions; they do not imply a source exercise, automatic feedback or completion record.
- Pathline analytical idea: write the invoice-line grain and source units, distinguish overlapping missing/negative/cancellation diagnostics from sequential disjoint cohort exclusions, and reconcile your aggregate row counts.
- Pathline analytical idea: define recency, distinct-invoice frequency and retained positive-sales monetary with an explicit cutoff and tie-preserving scoring policy. State the denominators and policy limitations.
- Pathline analytical idea: frame a commercial action as a historical-data hypothesis with a future measurement plan, rather than as an observed effect.
Limitations to check
- The confirmed explanation concerns dataset metadata and the variable dictionary, not a provider RFM lesson or assignment. The source dataset is confirmed; provider exercises, feedback and executable practice are not confirmed.
- RFM policy, objectives, prerequisites, preparation and the 180-minute duration are Pathline research analysis. Duration is an unmeasured researcher estimate, not a provider duration or demonstrated learner outcome.
- The transactions cover 2010-12-01 to 2011-12-09 at one UK retailer. Source timestamps have no established timezone; no current Hong Kong-market inference follows.
- Retained positive-sales monetary differs from net revenue, profit or retained customer value. Relative-cohort RFM scores and ties depend on the chosen policy; no unique provider answer or intervention effect is implied.
- Source no-missing metadata is separate from observed empty CSV fields. Numeric CustomerID and detailed history do not certify anonymity; no identity linkage or customer-level publication is proposed.
- Use UCI’s original download link and dataset-specific terms. Pathline does not supply raw data, internal code, customer-level outputs or a grader; copyright licence conditions do not settle all privacy or other rights.
Learning source and review reference
Source version: Dataset donated2015-11-05; API last_updated2024-10-21; actual CSV identified by SHA256 · Unit ER-20261013-01-WP05-U0005, version 2 · Review ER-20261014-A0-R2-FINAL-001