🛍️ Retail Store Sales Forensics

Overview

Public dataset from Kaggle's Rossmann Store Sales competition (2013–2015): daily sales and customer counts from 1,115 German drugstores.

Data Structure

Table	Records	What It Captures
Daily Sales	~1M rows	Sales amount, customer count per store per day
Store Metadata	1,115	Store type, assortment level, competition distance, promo participation

Causal Reasoning Model

Nested causal chains trace low sales through multiple diagnostic dimensions:

• Customer Footfall Drop: fewer visitors → lower sales, traced to day-of-week, holidays, school holidays
• Nearby Competitor: competition proximity + opening date → market share erosion
• Store-Type Sensitivity: different store types respond differently to promos and competition
• Anomaly Detection: per-store z-score normalization accounts for each store's baseline

Sample Questions

• "Why were sales low at store 85 on 2015-06-01?"
• "How did the nearby competitor affect this store's sales?"
• "What store-type patterns explain the promo response?"