MODULE 04 — LEARNER WORKSHEET Keep source data unchanged. Your files stay on your device. 1. Define the extract and build a useful data profile Download orders_raw.csv, orders_verified.csv and the setup/data dictionary. Without editing the raw file, record the reporting population, business key, currency, row count and distinct order count. List at least four checks you will perform before calculating sales. Then profile the verified file by month and status. Deliverable: a one-page profile containing raw records, distinct identifiers, duplicate groups, verified orders, completed orders and a paragraph stating the evidence boundary. Explain why the latest observed date cannot establish coverage of all real-world activity. My method / result / checks / remaining questions: 2. Turn data-quality findings into explicit rules Use quality_log_template.csv to write one issue statement for each of the eight listed identifiers. Add a rule, effect on sales/channel/delivery analysis, and the evidence needed to act. Separately record the count of genuine missing delivery observations after preparation. Distinguish a deterministic formatting fix from a value requiring authoritative confirmation. Deliverable: an eight-entry quality log plus three prioritised questions for a source owner. Do not delete all flagged records or invent an accuracy percentage from the number of passing rules. My method / result / checks / remaining questions: 3. Prepare data with an auditable correction log Create your own prepared copy from orders_raw.csv. Complete the eight log entries, applying only dictionary rules or exact reference-backed corrections. Resolve both duplicate groups and preserve genuinely blank delivery fields. Reconcile the raw, unique-ID, prepared and completed populations, then compare your prepared fields to orders_verified.csv using order_id. Deliverable: a prepared CSV, completed quality log and a record bridge. As an extension, describe what your process should do if a third Y007 row arrives with another channel and no source timestamp. My method / result / checks / remaining questions: 4. Explore distributions without deleting inconvenient values Filter the verified data to completed orders. Compute count, total, mean, median, Q1, Q3, IQR and both fences using inclusive interpolation. Build a frequency table using [0,100), [100,150), [150,200), [200,300) and [300,1100] EUR bins; the last interval includes its upper boundary. Identify flagged orders by ID and check their values against the reference. Deliverable: a distribution summary and a labelled baseline/exclusion comparison. Explain why the verified €1,000 order remains in the baseline even though it exceeds the fence. You may use a spreadsheet or the optional worked Python file after attempting the task. My method / result / checks / remaining questions: 5. Explain an aggregate trend using segments and weights Build a month-by-channel table with completed-order count, total value, AOV and order share. Reconcile each monthly count and value to your baseline. Reconstruct overall AOV from channel weights, then calculate an explicitly hypothetical 50/50 fixed-mix average for each month. Deliverable: a reconciled table, two weighted-average equations and a short explanation of how overall AOV can rise while both channel AOVs fall. Include April store’s sample size and high-value order. Explain why the scenario is neither a new actual total nor evidence of a causal marketing effect. My method / result / checks / remaining questions: 6. Keep missingness, coverage and denominators visible For each month, calculate eligible completed orders, observed delivery counts, missing counts, coverage, observed duration sum and observed mean. Demonstrate the effect of incorrectly filling missing durations with zero, keeping that example clearly labelled as wrong. Then calculate returned share using both all-status and completed-plus-returned denominators. Deliverable: a coverage table, two accurately named status shares and a sentence explaining why no all-completed-order delivery improvement is established. List the IDs with missing completed delivery and inspect their channel distribution. My method / result / checks / remaining questions: 7. Deliver a reproducible exploratory evidence pack Complete the evidence pack: raw and prepared data, quality log, scope note, row bridge, monthly metrics, reconciled channel table, distribution chart, coverage table and a 120–180 word decision note. Include one labelled sensitivity comparison and identify a next investigation. Use the ten-point rubric below to review your own work. Rubric: preparation/provenance 0–2; population and calculations 0–2; segment weights and sensitivity 0–2; missingness and denominators 0–2; communication and reproducibility 0–2. Award two when complete and correct, one when present but incomplete, and zero when missing or materially wrong. Aim for eight or more, with no zero in calculations or provenance. Keep unsupported causal claims out of the final note. My method / result / checks / remaining questions: No payment, registration or upload is needed for this public practice.