Manny-hub/RetailPulse-Analytics

RetailPulse Analytics operates 45 stores and an online marketplace in Nigeria, Ghana, and Kenya. With 150k monthly customers and 8k daily transactions, they need a unified source of truth. Centralizing this multi-channel data is essential for scaling operations and enabling consistent, data-driven decision-making. # RetailPulse Analytics Synthetic Data Generator Synthetic data generator for the three messy source systems described in your RetailPulse brief: - Inventory API JSON feed with inconsistent field naming, duplicates, timeout/rate-limit style metadata, and noisy stock data - Vendor shipment CSV files from multiple vendors, each with its own quirks and parsing issues - PostgreSQL-style order, customer, and order-item data with nulls, duplicates, and orphaned references ## What this generates ```text output/ api/ inventory_payload.json inventory_failures.json vendors/ vendor_001_nairobi.csv vendor_002_accra.csv ... postgres/ customers.csv orders.csv order_items.csv schema.sql seed_from_csv.sql ``` ## Quick start ```bash cd retailpulse_synth python generate_all.py ``` Optional arguments: ```bash python generate_all.py \ --vendors 12 \ --vendor-rows-min 40 \ --vendor-rows-max 120 \ --inventory-records 800 \ --customers 1500 \ --orders 4500 ``` ## Notes - The generator intentionally creates dirty data. - Vendor CSV files are separate by design so you can merge them yourself later. - PostgreSQL output is delivered as CSV plus schema/seed SQL helper files. - The API failures file simulates rate-limit and timeout style issues without requiring a real HTTP server. ## Main chaos patterns included ### Inventory API - `product_id` vs `sku` vs `productId` - `stock_quantity` vs `stock_qty` vs `qty_on_hand` - duplicated inventory records - negative or mismatched stock edge cases - timestamp field name drift - failure events for `429`, `408`, and `504` ### Vendor CSV - multiple date formats - different column names per vendor - product naming variations - duplicated rows - null fields - special characters / commas / quotes / ampersands / unicode - inconsistent quantity and unit-price formatting ### PostgreSQL order data - customer duplicates - nullable email / phone / shipping fields - orphaned `order_items` pointing to missing inventory SKUs or missing order IDs - statu …