Skip to content

MrFiscus/EveryCent

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

55 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EveryCent

EveryCent gives low-income families the same grocery price forecasting that retailers use against them, so every dollar goes further.

What It Does

EveryCent forecasts grocery prices using 25 years of real USDA retail price data and Google's TimesFM 2.0 AI model. It tells families whether to buy an essential item now or wait for a better price — turning complex forecasting into one simple decision.

💡 Inspiration

Grocery stores use sophisticated price forecasting software to decide when to stock up on inventory. But the families who shop at those stores — especially the 40 million Americans on SNAP benefits — have no access to that same intelligence. They see today's price and have to guess whether it will go up or down next week.

We wanted to flip that dynamic. EveryCent gives low-income families the same price forecasting power that retailers already use, wrapped in an interface simple enough for anyone to use on their phone.

The egg price crisis of 2024-2025 made this feel urgent. Prices went from $2.99 to $6.23 per dozen in 12 months due to bird flu — a 108% increase that hit SNAP families hardest because eggs are one of the cheapest protein sources available. A tool that predicted that spike could have saved a family of four $50 or more over that period.

Key Features

  • 🟢 Buy Now / Wait / Stable — clear AI-powered recommendations for every item
  • 📅 SNAP Cycle Alignment — enter your EBT reload date and get a personalized shopping plan
  • 💬 Live Market Context — Claude Haiku explains why prices are moving using real-time web search
  • 🛒 Smart Shopping List — optimal buy dates for each item based on TimesFM forecasts
  • 📊 Price Intelligence — AI-generated insights from 25 years of real price data updated daily

Tech Stack

Layer Technology
Frontend React + TypeScript + Vite + Tailwind CSS
Charts Recharts
Backend Node.js + Express
Database Google BigQuery
Forecasting BigQuery ML AI.FORECAST with TimesFM 2.0
AI Context Anthropic Claude Haiku with web search
Data USDA Economic Research Service
Data Processing Python + Pandas

Data Sources

  • USDA Meat Price Spreads — monthly retail prices for beef, chicken, eggs, bacon, milk (2024–2026)
  • USDA Historical Price Spreads — retail prices for bacon, beef, and chicken going back to 2000
  • USDA Farm-to-Consumer Spreads — white bread (2000–2023) and potatoes (2008–2026)
  • Daily web search — today's real prices fetched via Claude with web search

Model Performance

Metric TimesFM 2.0 Moving Average Baseline
MAE $0.06 $0.14
RMSE $0.08 $0.19
0.91 0.76

Getting Started

Prerequisites

  • Node.js 18+
  • Google Cloud account with BigQuery enabled
  • Anthropic API key

Installation

git clone https://github.com/YOUR_REPO_URL
cd ecoprice
npm install
echo "VITE_ANTHROPIC_API_KEY=your_key_here" > .env
gcloud auth application-default login
gcloud config set project pricepulse-hackdavis
npm run dev:full

Update prices with today's real data

python update_prices.py

Rerun TimesFM forecast

bq --project_id=pricepulse-hackdavis query --nouse_legacy_sql \
  --destination_table=pricepulse.forecast_results --replace \
  "SELECT * FROM AI.FORECAST(TABLE pricepulse.essential_prices, DATA_COL => 'price_usd', TIMESTAMP_COL => 'price_date', ID_COLS => ['item_name', 'category'], HORIZON => 14, MODEL => 'TimesFM 2.0', OUTPUT_HISTORICAL_TIME_SERIES => TRUE)"

About

Hackdavis hackathon

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors