You can use official kaggle-api client (which is already pre-installed in all our environments) to download any Kaggle dataset directly to a FloydHub Workspace. You'll need to add your Kaggle API key to each Workspace on FloydHub in order to use the kaggle-api. This guide walks through the steps:
- From your Kaggle account (https://www.kaggle.com/<username>/account) you need to create a token ( a small JSON file with contents that look something like this:
{"username":"your_kaggle_username","key":"your_kaggle_api_key"})and set the respective field as environment variables inside your Workspace Terminal by running these commands:
$ export KAGGLE_USERNAME=your_kaggle_username
$ export KAGGLE_KEY=your_kaggle_api_key- That's it! Your Kaggle API is now configured, and you're ready to use the
kaggle-apicommand line tool. You can now download the dataset using thekagglecommand described in the dataset page:
- You will find the dataset in zip archive in the
/root/.kagglefolder. Feel free to move this new dataset back to your/floyd/homedirectory if you'd like to persist it between Workspace sessions. - Pro tip -- you can also upload the Kaggle dataset as a FloydHub dataset from the Workspace by following these steps.