Kaggle API Settings

Introduction

This post describes the setup steps required before using Kaggle API. These settins are required to execute Kaggle API outside the Kaggle Kernel pllatform, e.g. in Colab or in a private computing platform. When executing a notebook inside Kaggle Kernel platform, the described settings are not needed. Issue is presented here.

Kaggle API can be so useful, and save much efforts that would otherwise be required, e.g. consuming Kaggle datasets is trivial with the API, but would have required a pre-arrangement of private storage for the dataset - in a local pc or in the cloud, according to executiion mode - otherwise.

Before one can interact with public Kaggle API, 2 actions should be taken:

• Package Instalation
• Authentication

Next is a detailed description of these 2 steps followed by some API usage examples.

BTW - You should have a Kaggle account for that - sign up in case you are not there already!

Kaggle API Package Installation

Simply use pip (package installaerfor python):

pip install kaggle

Authentication

To authenticate, one should do 2 actions:

1. Generate a token file - as detailed next.
2. Store the token file under

   ~/.kaggle
   

   in linux/osx or under

   C:\Users<Windows-username>.kaggle\ 
   

Here are details instructions for doing that:

Generate a token file

1. Login to your Kaggle Account
2. In Kaggle web page click: user profile picture (upper right of page) -> Account -> Create New API Token
3. The kaggle.json token file should be downloaded to your local storage.

Store the token

1. Create a directory:

     mkdir ~/.kaggle
   

2. Copy token file to the created directory:

     cp ~/Downloads/kaggle.json  ~/.kaggle
   

   Beware! the token file is now exposed, so you might want to change it’s permissions:

chmod 600 ~/.kaggle/kaggle.json

Notes on setting the token in Colab

If running in Colab, then the token file should be uploaded to the platform. Here below is a section of code that should be addd to the Colab notebook. This section provides an iteractive file upload button for uploading kaggle.json.

from google.colab import files

uploaded = files.upload()
  
# Move kaggle.json into ~/.kaggle:
!mkdir -p ~/.kaggle/ && mv kaggle.json ~/.kaggle/ && chmod 600 ~/.kaggle/kaggle.json

Now Kaggle API is ready to go.

Kaggle API Usage Examples

(In the below comands the extrac exclamation sign prefix is needed to run commands from within a jupyter notebook).

• List Kaggle datasets:

!kaggle datasets list

• Doenload a dataset:

!kaggle datasets download -d coloradokb/dandelionimages

That’s it. Enjoy!