Welcome to fif_recsys¶

Ranking Brazilian fixed income funds.

Start Here!¶

See the Notebooks > E2E Example tab in the docs on a full minimal recsys example on brazilian fixed income funds.

To automate this as a scheduled job, we have a Dockerfile that runs the full pipeline based on a config YAML. In a production scenario, we would have it as an Airflow job and evalute which steps should be splitted into specific airflow tasks. But for now, a cron job that executes the full pipeline as a whole is enough, refer to the manifest.yaml for the config YAML.

To run it as a docker container, run the commands:

make docker-image

docker run --rm -v "$(pwd)/manifest.yaml:/manifest.yaml" -v "/tmp/fif_data:/data" acli pipeline /manifest.yaml /data

Installation¶

pip install fif_recsys

Usage¶

The CLI provides a set of commands:

fif data --help
fif feature --help
fif model --help
fif policy --help

Notebooks¶

Interactive examples are available as Jupyter notebooks and are converted to Markdown for inclusion in the site. To regenerate the converted docs locally run:

make convert-notebooks

Example notebook: notebooks/example.md

Development¶

For developers who want to extend this CLI or use it as a template:

Testing¶

Run the test suite:

pytest tests/ -v

Building Documentation¶

Build the documentation locally:

mkdocs serve -f docs/mkdocs.yml

The CLI is built with: - Typer - Modern CLI framework - Rich - Rich text and beautiful formatting - Pytest - Testing framework