Restaurant Stats

Problem

My dining spending was scattered across four financial sources (Bank of America, Chase, Capital One, and Venmo), each with its own PDF statement format. I wanted a single view of where I actually eat: how often, how much, and on which card, without uploading years of financial statements to some third party.

Approach

A three-stage Python pipeline (extract to build to dashboard):

• Extract ingests raw credit-card and Venmo PDF statements.
• Build detects and canonicalizes dining transactions across the four different statement formats.
• Dashboard produces per-restaurant analytics.

Financial input data stays fully local and git-ignored.

Highlights

• Handles four distinct statement formats and reconciles them into one canonical set of dining transactions.
• Computes per-restaurant stats: number of visits, total spend, average ticket size, first and last visit dates, and which cards were used where.
• Keeps all financial input local and out of version control.