Behavior events
20.69M
Raw event rows across the source files
Selected work / 12 · Retail analytics
ATLAS BI
A source-grounded preview of an end-to-end retail intelligence platform: data engineering, business analytics, and four machine-learning workflows.
INTERACTIVE DATA PREVIEWVerified source row counts · not distinct orders · repository private for now
Purchase events
1.287M
Purchase-event rows; not confirmed orders
Coverage
5 months
October 2019 through February 2020
Dataset volume
Monthly source records
Counts are from the original monthly files.
4,102,283
Oct ’19
4,635,837
Nov ’19
3,533,286
Dec ’19
4,264,752
Jan ’20
4,156,682
Feb ’20
Dataset: REES46 eCommerce Events History in Cosmetics Shop. The source does not provide order IDs, costs, returns, stock levels, or customer locations. The preview therefore avoids claiming profit, distinct order totals, inventory optimization, or geographic results.
Data pipeline
Chunked validation, quality checks, Parquet, DuckDB and reproducible lineage.
Business analytics
Sales, product, customer and session-based basket views through a secured API.
Predictive models
Forecasting, inactivity-proxy classification, customer segments and anomaly triage.
Responsible insights
Validated question intents and explicit limitations; no unrestricted generated SQL.