Avocado

Avocado turns city data, real-time news, weather, & a weighted ML model into a simple AvoScore, so you can quickly see how affordable a city really is, whether it fits your life, not just your budget.

Comment

Avocado 🥑 — plan your next move

Avocado is an affordability intelligence engine that helps users understand the true cost of living in any major city. Designed for students, newcomers, families, and relocating professionals, Avocado provides a complete data-driven breakdown of all major lifestyle expenses and city conditions.

The platform analyzes real-time data, economic indicators, news sentiment, and a custom weighted ML model to generate the AvoScore, a 0–10 affordability index that reflects how financially livable a city is.

AvoScore Legend

  • 3 → Low cost
  • 4–7 → Mid cost
  • 8+ → Expensive

Live demo: https://avocado.amanpurohit.com

What Avocado Delivers

Avocado provides a consolidated, real-time snapshot of a city’s livability, including:

  • Cost-of-living data across rent, groceries, dining, and transit
  • Purchasing power and economic stability
  • Crime, safety, transportation, and finance-related news
  • Live weather and environmental conditions
  • City sentiment computed from real-time news feeds
  • A machine-learning powered affordability score (AvoScore)
  • Gemini-powered conversational insights for interpretation

The goal is simple:
Help users make informed, data-backed decisions about where to live.

AvoScore — The Affordability Engine

The AvoScore is produced through a hybrid weighted ML model that combines classification, regression, and sentiment analysis to deliver a stable, interpretable affordability measure.

Feature Inputs

Category Features
Housing Rent index, price-to-income ratio
Groceries Grocery index
Dining Restaurant index
Safety Crime index, safe-walking-at-night score
Economics Local purchasing power, employment volatility
Transport Transit accessibility and cost
Weather Climate & comfort score
Sentiment VADER sentiment from real-time news

EDA & Preprocessing

Libraries and methods used:

  • pandas — data ingestion, merging, cleanup
  • numpy — numerical transformations
  • matplotlib, seaborn — visualization and correlation analysis
  • scikit-learn — scaling, preprocessing, and baselines
  • scipy — outlier detection (IQR trimming)
  • NLTK (VADER) — sentiment analysis
  • pycountry, geopy — city validation and coordinate mapping

Modeling Approach

Avocado uses a two-stage ML pipeline:

1. Logistic Regression (Classification Baseline)

Predicts affordability tiers:

  • Low
  • Medium
  • High

2. XGBoost Regressor

Generates a continuous affordability score.

Weighted Ensemble Formula

AvoScore =
  (0.70 * XGBoost) +
  (0.15 * city_sentiment) +
  (0.10 * weather_comfort) +
  (0.05 * local_purchase_power)

This weighted method provides balanced predictions across diverse cities, smoother scoring, and improved generalization.

Sample Classification Performance


      precision    recall  f1-score   support

    Low Cost       0.89      1.00      0.94        16
 Medium Cost       0.89      0.80      0.84        10
   High Cost       1.00      0.83      0.91         6

    accuracy                           0.91        32
   macro avg       0.93      0.88      0.90        32
weighted avg       0.91      0.91      0.90        32

System Architecture

                         AVOCADO SYSTEM ARCHITECTURE
                         ----------------------------------

                          ┌─────────────────────────────────┐
                          │          Frontend (Vercel)      │
                          │        Next.js + TypeScript UI   │
                          │  - City search                   │
                          │  - City detail pages             │
                          │  - AvoScore visualizations       │
                          └───────────────┬─────────────────┘
                                          │  HTTPS Fetch
                                          ▼
                   ┌────────────────────────────────────────────────────┐
                   │                 Backend (Cloud Run)                │
                   │       Python FastAPI in Docker Container           │
                   │             Auto-scaled via Cloud Run             │
                   ├───────────────┬──────────────────────┬────────────┤
                   │               │                      │            │
     ┌─────────────▼───────┐ ┌────▼────────────────┐ ┌────▼───────────────┐
     │  Weather Service     │ │  News Aggregation    │ │ ML Affordability    │
     │  WeatherAPI.com      │ │ NewsData / Currents  │ │ Engine (AvoScore)   │
     │                      │ │ Crime/Finance/       │ │ Weighted Model      │
     │                      │ │ Transport/Events     │ │ XGBoost + LR        │
     └──────────────────────┘ └──────────────────────┘ └────────┬───────────┘
                                                                 │
                                                  ┌──────────────▼──────────────┐
                                                  │       AvoScore API           │
                                                  │    0–10 Affordability Index  │
                                                  │        + Explanations        │
                                                  └──────────────────────────────┘

AI Assistant (Gemini Integration)

Avocado uses Gemini to provide conversational explanations, comparisons, and real-time insights about:

  • Affordability breakdowns
  • City-to-city comparisons
  • Risk and volatility (economic or environmental)
  • Forecast trends and city outlook

This turns data into interpretable, user-friendly guidance.

Why Avocado?

Affordability influences every aspect of daily life — housing, transportation, food, lifestyle, and even small decisions like whether you can buy an avocado. Avocado makes the affordability question clear, actionable, and data-driven.

It delivers financial transparency for anyone making a major life decision about where to live.

What's Next?

We’re expanding Avocado beyond major metropolitan hubs to support real-time affordability insights for any city, town, or region. That means: Ingesting more granular, real-time data sources (regional rents, local transit, groceries, utilities). Scaling the AvoScore model to handle thousands of locations with dynamic retraining.

Factoring in live shocks like policy changes, climate events, or economic swings, to show how affordability shifts over time.

Built With

Share this project:

Updates