training-song 0.2.2

Audio Motivation for Data Scientists and ML Engineers

Training Song

Description

Plays a Billboard Number 1 song corresponding to how accurate your ML model is.

For example if your model is 95.5% accurate then you will hear the number 1 song from 50% through 1995 (Vogue by Madonna 👑).

Take your metrics from A Hard Day's Night (64%) to Mo Money Mo Problems (97%).

How to use

Once you've trained your model, simply wrap your metric in ts(..) as follows:

from trainingsong import ts

model.fit(X_train, y_train)
y_pred = model.predict(X_test)

accuracy = ts(accuracy_score(y_test, y_pred))

>> Congrats your model got an accuracy of 92 percent!
>> The Number 1 song 92.0% through the 1900s on the hot-100 chart was
   Black Or White by Michael Jackson.
>> The date was 1992-01-01 and the song was on the chart for 7 weeks.

Installation

Use the package manager pip to install trainingsong.

pip install training-song

Local Development

The API docs can be found here

You can install the development dependencies with:

poetry install

And you can run the tests using

poetry run pytest

Before committing, please run the following to run the tests:

tox

It's recommended to use uvicorn to run the server locally, which is installed as a dependency.

Please create a Postgres database and set the DATABASE_URL as an environment variable in the .env file. The db.py file defines the schema and gives a function to create the table.

Additionally if you're editing the main API then you will need to create a Spotify app and set include the CLIENT_ID and CLIENT_SECRET as environment variables. In this case you will also need to setup a Vercel account and deploy the API to it. Then you can use the Vercel CLI to run the server locally.

vercel .

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

License

MIT

Acknowledgements

Thanks to Spotify for the API and Billboard for the data.