Viewing a saved revision (read-only)
eefluKnT89lOBt1oitq1l
WWhyoumust
Last edited Nov 12, 2026
Created on Nov 12, 2026
Created from react-d3-starter template
Olist Brazilian E-Commerce Datasets — Documentation
Datasets
1. olist_customers_dataset.csv
- Rows: 99,441 · Columns: 5
- Apparent subject: One row per customer order instance; identifies customers and their geographic location.
| Column | Type (Munzner) | Meaning | Missing |
|---|---|---|---|
| customer_id | Nominal (key) | Identifier per order-customer row | 0 |
| customer_unique_id | Nominal (key) | Stable customer identity across orders | 0 |
| customer_zip_code_prefix | Quantitative / Geographic identifier | Brazilian 5-digit ZIP prefix | 0 |
| customer_city | Nominal | City name (lowercase) | 0 |
| customer_state | Nominal | Brazilian state abbreviation (26 values) | 0 |
2. olist_order_payments_dataset.csv
- Rows: 103,886 · Columns: 5
- Apparent subject: Payment records; an order may have multiple sequential payments (rows > orders).
| Column | Type | Meaning | Missing |
|---|---|---|---|
| order_id | Nominal (key) | Order identifier | 0 |
| payment_sequential | Ordinal | Payment attempt number within the order (1–8) | 0 |
| payment_type | Nominal | Method: credit_card, boleto, voucher, debit_card (4 values) | 0 |
| payment_installments | Ordinal | Number of installments (1–13) | 0 |
| payment_value | Quantitative | Amount in BRL (0.01–2288.31 in sampled profile) | 0 |
3. olist_products_dataset.csv
- Rows: 32,951 · Columns: 9
- Apparent subject: Product master data with physical dimensions and Portuguese category names.
| Column | Type | Meaning | Missing |
|---|---|---|---|
| product_id | Nominal (key) | Product identifier | 0 |
| product_category_name | Nominal | Portuguese category (58 observed) | 29 |
| product_name_lenght | Quantitative | Character count of product name | 29 |
| product_description_lenght | Quantitative | Character count of description | 29 |
| product_photos_qty | Ordinal (numeric) | Number of photos (1–17) | 29 |
| product_weight_g | Quantitative | Weight in grams | 0 |
| product_length_cm / product_height_cm / product_width_cm | Quantitative | Dimensions in cm | 0 |
4. olist_sellers_dataset.csv
- Rows: 3,095 · Columns: 4
- Apparent subject: Seller registry with location.
| Column | Type | Meaning | Missing |
|---|---|---|---|
| seller_id | Nominal (key) | Seller identifier | 0 |
| seller_zip_code_prefix | Quantitative / Geographic identifier | ZIP prefix | 0 |
| seller_city | Nominal | City name | 0 |
| seller_state | Nominal | State abbreviation (19 observed) | 0 |
5. product_category_name_translation.csv
- Rows: 71 · Columns: 2
- Apparent subject: Lookup table mapping Portuguese category names to English.
| Column | Type | Meaning |
|---|---|---|
| product_category_name | Nominal (key) | Portuguese category |
| product_category_name_english | Nominal | English translation |
Relationships (candidate join keys — unverified, use with caution)
order_payments.order_id↔ orders-level datasets (an orders dataset is not present in this project, so this key cannot be joined within the provided files alone).products.product_category_name↔translation.product_category_name(71 translation rows vs 58 observed product categories — translation is a superset; products may have categories absent here, and missing values in products map to nothing). Defensible join, many-to-one, at product grain.customers.customer_zip_code_prefix↔sellers.seller_zip_code_prefix: shared ZIP-prefix semantics but different entity grains (customers vs sellers); a join would express co-location, not an entity relationship. Not used.- Not provided (would be needed for full analysis): orders, order_items, order_reviews, geolocation datasets.
Visualization
Type: Horizontal bar chart — total payment value (BRL) by payment type, from olist_order_payments_dataset.csv (single dataset, no risky joins).
- Encodings: y = payment_type (nominal, alphabetical); x = sum of payment_value (quantitative); color = single accessible categorical hue; value labels at bar ends.
- Interactions: tooltips on hover showing total and share; responsive sizing via container dimensions.
- Grain note: payments dataset is at payment-row grain (multiple rows per order); we aggregate with
d3.rollup, which is valid regardless of order multiplicity. - Rationale: directly answers "how is payment value distributed across methods" with real, verified fields only.
Implementation notes
- CSV parsed with
d3.csv(RFC4180-compliant, handles quoted embedded newlines); header is the first parsed record. - Numeric parsing coerces
payment_value; invalid/missing values are excluded from sums. - Data loaded at runtime from
/data/olist_order_payments_dataset.csv; nothing hardcoded.