Walter Mellon's Marketplace: Reconcile Ten Stores at Scale

The challenge

Bring ten stores' sales into one dataset. Fix the CSV formatting, keep each store's IDs straight, and find out why the busiest store isn't always the one with the most revenue.

Start here

Open the SharePoint folder linked above. store1_transactions.csv through store10_transactions.csv hold the records, totaling about 1.77 GB. WalterMellon_1_receipts.pdf through WalterMellon_10_receipts.pdf are compact generator summaries with clipped receipt previews, not full receipt books. There is no complete JSON answer export in this folder.

Inspect the first few raw CSV lines and each store's summary PDF. Every CSV has transaction_id, customer_id, customer_name, date, time, items, and total_price. Store identity is included by the filename, and transaction numbering restarts in each file.

Try it

1. Recover a valid seven-field record structure despite unescaped JSON quotes in the items column.

2. Add store_id and create transaction/item keys that remain unique across all ten stores.

3. Reconcile each store's transaction count and revenue with its PDF summary.

4. Rank stores by transaction count, total revenue, and average basket value, and explain the different winners.

What to build

Build a repair and validation walkthrough, normalized schemas, per-store summary results, and a combined reconciliation. Record rejected rows and state which checks ran over all rows versus a sample.