lschiemanowski commited on
Commit
77aef73
·
verified ·
1 Parent(s): 9885d01

Add Online Retail II DuckDB description

Browse files
Files changed (1) hide show
  1. online-retail-ii/1.0.0/DATABASE.md +34 -0
online-retail-ii/1.0.0/DATABASE.md ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ This DuckDB contains the **Online Retail II** transaction-line dataset from the UCI
2
+ Machine Learning Repository. The source describes a UK-based non-store retailer, and
3
+ the records cover **1 December 2009 through 9 December 2011**. Unit prices are in GBP.
4
+
5
+ For ordinary analysis, use `analysis.transaction_lines_nonoverlapping`. It contains
6
+ 1,044,848 product lines and removes 22,523 rows where the two source worksheets overlap,
7
+ while intentionally retaining duplicate occurrences within an individual worksheet.
8
+ Each row represents one product line on an invoice, not one complete invoice.
9
+
10
+ Important analytical conventions:
11
+
12
+ - `quantity * unit_price_gbp` is the signed line value. Negative quantities therefore
13
+ reduce net sales rather than counting as positive purchases.
14
+ - Returns and cancellations can have negative quantities, and cancellation invoice IDs
15
+ commonly begin with `C`.
16
+ - `invoice_ts` is the invoice timestamp and has no timezone attached.
17
+ - `customer_id` and `description` can be missing. A small number of unit prices are zero
18
+ or negative.
19
+ - `country` is the invoice country. The data contains 43 distinct country values.
20
+
21
+ The DuckDB also retains both original worksheets in `raw.transaction_lines`. The
22
+ `metadata.dataset` and `metadata.quality_summary` relations contain provenance, row
23
+ counts, and documented quality conditions. Prefer the `analysis` view unless a question
24
+ explicitly concerns the raw overlap or source-sheet provenance.
25
+
26
+ ### Example questions
27
+
28
+ - How did monthly net sales change during 2011?
29
+ - Which countries outside the United Kingdom generated the most net sales?
30
+ - Which products had the largest cancellation or return quantities?
31
+ - How many customers purchased in consecutive calendar months?
32
+
33
+ Source: [UCI Online Retail II](https://doi.org/10.24432/C5CG6D). The packaged database
34
+ and this description are distributed under CC BY 4.0.