SauceBros Sales Analysis

Inspiration

What drove you to build this?
– Our love of pizza and the challenge of optimizing operations for a fast‑growing local chain.
– The need to reduce food waste and labor cost by better forecasting demand.

What it does

Summarize the core features in 2–3 bullets:

Demand Forecasting: Predicts weekly and monthly pizza sales using historical data and seasonal trends.
External Data Integration: Pulls in local event calendars and consumer‑spending indices to improve accuracy.
Dashboard Visualization: Interactive charts show inventory recommendations and staffing needs.

How we built it

Outline your tech stack and key steps:

Data Collection: Loaded and cleaned 3 years of Sauce Bros’ POS data via Pandas.
Feature Engineering: Extracted month‑day trends, holiday flags, and local event indicators.
Modeling: Tested SARIMA, Prophet, and XGBoost regressors with cross‑validation.
Visualization: Built an interactive dashboard in React with Recharts.

Challenges we ran into

Be candid—what slowed you down or required creative problem‑solving?

Handling year‑rollover when dates only included “Apr 30” style month/day strings.
Incorporating one‑off local events (e.g. high‑school football games) without overfitting.
Deploying the model pipeline so it could retrain automatically each month.

Accomplishments that we're proud of

Highlight wins, big or small:

**XG boost with 94% accuracy
**Optimize employee schedule
**Integrated visualizations into a web app

What we learned

Reflect on skills, tools, or domain knowledge gained:

The importance of embedding business insight (promo dates, staffing constraints) into ML pipelines.
How to tune SARIMA parameters to capture both seasonality and trend.
Best practices for cross‑functional collaboration—bringing operations and finance into model reviews.

What's next for SauceBros Sales Analysis

Lay out your roadmap:

🔄 Automate daily data ingestion and model retraining on AWS Lambda.
📈 Add competitor‑benchmarking by scraping Yelp and Google Trends.
🤖 Experiment with reinforcement learning to optimize dynamic pricing during peak hours.