Crowdless

Minimal pipeline to compute and visualize a crowdness score per area over time. Includes a TypeScript UI demo and a Jupyter notebook for data processing, exploration, and scoring.

---

Contents

• UI_Demo/ — lightweight web UI demo for browsing areas and inspecting scores
• Crowdness_API_analysis_Demo.ipynb — generic, feature-based scoring notebook

---

Data Format

Place per–area JSON files under areas_output/:

{
  "area": "Syntagma",
  "items": [
    {
      "datetime": "2025-07-01T13:00:00Z",
      "temperature_celsius": 30.5,
      "relative_humidity_percent": 55.0
    }
  ]
}

UI Demo

Configure the demo’s data source to point to your precomputed outputs or service as needed.

---

Notebooks

Generic Feature-Based Scoring

File: Crowdness_API_analysis_Demo.ipynb

Capabilities:

• Reads areas_output/*.json
• Builds a lookup dataframe areas_output/metrics_database.pkl
• Saves a simple, tunable scoring model areas_output/crowdness_model.json
• Single and batch predictions via nearest-timestamp lookup

Data Collector Notebook

File: Crowdness_API_analysis_Demo.ipynb

Adds a cell that writes collect_s2_catalog.py (Copernicus Data Space, Sentinel-2 L2A), collects scene metadata for predefined areas/date ranges, normalizes items, and saves JSON under areas_output/ (per-area or per-day).

Example:

python collect_s2_catalog.py --areas Syntagma,Kallithea --start 2025-01-01T00:00:00Z --end 2025-01-07T23:59:59Z --cloud-max 30 --out areas_output

Place JSON data under areas_output/, open the notebook(s), and run cells top-to-bottom. Edit FIELD_MAP in the generic notebook to match your numeric fields.

---

Scoring Model (Notebook)

Interpretable, rule-based score:

• feature_x: triangular peak at ideal_x, falling to 0 at ideal_x ± range_x
• feature_y: full score at or below ideal_y, linear penalty above up to 0 at ideal_y + range_y
• Final score: weighted average of the two sub-scores in [0, 100]

Tune crowdness_model.json parameters to fit your use case.

---