crunchdao's Introduction

Python API for the Crunchdao machine learning tournament

Interact with the Crunchdao tournament API using Python.

If you encounter a problem or have suggestions, feel free to open an issue.

Installation

pip install --upgrade crunchdao

Usage

Some actions (like uploading predictions) require an apikey to verify that it is really you interacting with Crunchdao. Keys can be passed to the Python module as a parameter or you can be set via the CRUNCHDAO_API_KEY environment variable

Example usage

import crunchdao
# some API calls do not require logging in
client = crunchdao.Client(apikey="foo")
# download current dataset
client.download_data(directory=".")
# get information about your submissions
submissions = client.submissions()
print(submissions)  # this is a pandas Dataframe
# get configure of the current dataset
client.dataset_config()
# upload predictions
predictions = ....  # pandas DataFrame containing your predictions
submission_id = client.upload(predictions)
# set comment for the submission, to remember which model that is etc
client.set_comment(submission_id, "Great model, learning_rate=0.01")
# Get your resolved targets live score on master dataset
client.get_scores()

Recommend Projects