TrabahoLens

TrabahoLens is a small research project for exploring Philippine occupations, employment size, wages, and generative-AI exposure in a static treemap view.

The repository is open-source-ready and runnable out of the box: the committed demo data is already built, so you can launch the site locally without any API key.

Quickstart

uv sync
uv run python scripts/serve.py

Then open http://127.0.0.1:8000.

Common Commands

# Run code + data checks before opening a PR
uv run python scripts/check.py

# Rebuild site/data.json from the committed derived CSV + score cache
uv run python scripts/build_site_data.py

# Rebuild the derived CSV from the stats JSON
uv run python scripts/make_csv_ph.py

# Optional: rescore occupations with OpenRouter, then rebuild the site data
cp .env.example .env
# add OPENROUTER_API_KEY to .env
uv run python scripts/score_ai_only.py
uv run python scripts/build_site_data.py

# Optional: score with a specific OpenRouter model
uv run python scripts/score_ai_only.py --model claude-3-5-haiku
uv run python scripts/score_ai_only.py --model google/gemini-2.5-flash
uv run python scripts/score_ai_only.py --model openai/gpt-4o-mini

Data Layout

• data/reference/occupations_seed.json Curated source/reference occupation records used to guide PH-specific shaping.
• data/derived/occupations_ph.csv Derived occupation table used by the supported scoring and site build steps.
• data/derived/scores_ai_only.json Derived AI exposure cache used by the site.
• site/data.json Built frontend dataset used by the static visualization.
• data/raw/ Ignored local-only fetch output for O*NET and PSA downloads.

Only the files needed for local viewing and optional rescoring are committed. Pipeline intermediates are regenerated locally and ignored.

Pipeline

The supported pipeline is:

uv run python scripts/fetch_onet.py
uv run python scripts/filter_ph_occupations.py
uv run python scripts/assign_ph_stats.py
uv run python scripts/make_csv_ph.py
uv run python scripts/score_ai_only.py   # optional, requires OPENROUTER_API_KEY
uv run python scripts/build_site_data.py

The repo keeps the derived demo artifacts committed so contributors can inspect the visualization without rerunning the whole pipeline.

Data Sources

• O*NET 30.2 Database for occupation descriptions and job-zone inputs
• Philippine Statistics Authority OpenSTAT for employment and wage reference material used during dataset construction

Notes

• OPENROUTER_API_KEY is only needed for rescoring.
• The default scorer model is claude-3-5-haiku.
• You can switch models through OpenRouter with --model, including Claude, Gemini, GPT, and other compatible providers.
• More detailed scoring notes live in docs/openrouter-setup.md.