GabrielGPT

README

3 min read

Coworking Space Explorer & AI Analyzer

Webpage

A Streamlit web app to discover, compare, and analyze coworking spaces using advanced filtering, clustering, and AI-powered scoring.

🚀 Features

  • Filter & Search:
    Find coworking spaces by city, price range, and must-have amenities.

  • Amenity Extraction:
    Parses amenities from both structured lists and free-text descriptions, removing duplicates and irrelevant items (like WiFi).

  • Price Normalization:
    Normalizes prices within each country or city to enable fair comparisons across different currencies and markets.

  • Recommendation System:
    Suggests similar coworking spaces based on amenity and price similarity using cosine similarity.

  • Workspace Clustering:
    Groups spaces into categories (e.g., Budget-Friendly, Premium, Creative Studios) using KMeans clustering for style-based exploration.

  • Top Rated Spaces:
    Displays and maps the best-rated coworking spaces by city, including price and rating metrics.

  • AI Scoring:

    • Trains a RandomForestRegressor on a synthetic quality score (combining amenities, price, and random noise for realism).
    • Predicts a quality score (1–5) for each space.
    • Provides a color-coded, emoji-enhanced score display.
    • Offers detailed breakdowns: price analysis (relative to local market), amenities analysis, competitive ranking, and percentile.
    • Generates actionable recommendations for improvement.

  • Live Model Training:
    Retrain and reload the AI model at any time with a button in the AI tab, using the latest data.

  • Explainability:
    The app provides transparent, user-friendly explanations and recommendations for each space.

🏗️ How It Works

  1. Data Loading:
    Loads coworking and amenities data from CSV files in /src/results/.

  2. Data Processing:

    • Cleans and merges datasets.
    • Extracts and enhances amenities.
    • Normalizes prices and encodes categorical variables.

  3. User Interaction:

    • Users filter and select spaces in the sidebar and main tabs.
    • Similar spaces and clusters are visualized and compared.
    • In the AI tab, users can retrain the model and analyze any space.

  4. AI Model:

    • Feature engineering includes total amenities, normalized price, city encoding, and top amenities.
    • Synthetic target score is generated for supervised learning.
    • Model is trained and evaluated live in the app.
    • Model and features are saved to /src/ai/ for persistence.

  5. Analysis & Recommendations:

    • Each space receives a detailed, explainable AI score.
    • The app provides pricing, amenity, and competitive analysis, plus actionable suggestions.

📦 File Structure

.
├── src
│   ├── ai
│   │   ├── model.pkl
│   │   └── features.pkl
│   ├── results
│   │   ├── extracted_amenities.csv
│   │   ├── merged_coworking_spaces.csv
│   │   └── MergedPlacesScoreDistance.csv
│   └── Images
│       ├── LocationMap.png
│       ├── CorrelationHeatmap.png
│       └── DataProcessing.png
├── app.py
├── requirements.txt
└── README.md

Future Improvements

  • User Preferences: Save user preferences for faster future searches
  • Additional Data Sources: Integrate more coworking space databases
  • Advanced Filtering: Add more granular filtering options like noise level and workspace type
  • Mobile Optimization: Improve responsive design for mobile users
  • Community Reviews: Incorporate user-generated feedback
  • Booking Integration: Enable direct space reservations

📌 Author: Gabriel Fernandes Pinheiro
🔗 LinkedIn | GitHub

Graph View