Kaggle Getting Started: Account, Keys & First Request (2026)

Getting started overview

Getting started with Kaggle involves creating a user account, generating API credentials, and making an initial programmatic request to confirm access. Kaggle provides an integrated environment for data exploration and model building, which includes cloud-based notebooks with access to GPUs and TPUs, eliminating the need for extensive local setup Kaggle API documentation. This process facilitates participation in data science competitions, sharing datasets, and collaborating on machine learning projects.

The core steps covered in this guide are:

1. Account Creation: Registering a new Kaggle user account.
2. API Key Generation: Obtaining an API token for authenticated access.
3. First Request: Using the API key to download a dataset programmatically.

Completing these steps prepares a user to interact with Kaggle's platform for various data science activities, from accessing datasets to submitting competition solutions.

Quick reference table

Create an account and get keys

To access Kaggle, you must first create a user account. This account serves as your identity on the platform and is required for all activities, including participating in competitions, utilizing notebooks, and accessing datasets.

Account registration

1. Navigate to the Kaggle registration page.
2. You can register using a Google account, which is recommended given Kaggle's ownership by Google, or with an email address Kaggle Homepage.
3. Follow the on-screen prompts to complete the registration process, including verifying your email address if you chose that option.
4. Once registered, log in to your new Kaggle account.

API key generation

For programmatic interaction with Kaggle, an API token is required. This token authenticates your requests, allowing you to download datasets, submit competition entries, and manage other resources via the Kaggle API client.

1. After logging in, navigate to your Kaggle Account Settings page.
2. Scroll down to the 'API' section.
3. Click the 'Create New API Token' button. This action downloads a file named kaggle.json to your browser's default download directory. This file contains your username and API key.
4. Security Note: The kaggle.json file contains sensitive credentials. Treat it like a password and keep it secure. Do not share it publicly or commit it to version control systems.

Install and configure the Kaggle API client

The official Kaggle API client is a Python package that simplifies interacting with the Kaggle platform. It can be installed using pip.

1. Install the client: Open your terminal or command prompt and run the following command:

   pip install kaggle

2. Configure API credentials: The kaggle.json file downloaded earlier needs to be placed in a specific directory for the API client to find it.
• Linux/macOS: Move the kaggle.json file to ~/.kaggle/. If the .kaggle directory does not exist, create it:

  mkdir -p ~/.kaggle
  mv /path/to/kaggle.json ~/.kaggle/
  chmod 600 ~/.kaggle/kaggle.json

  The chmod 600 command sets permissions to ensure only the owner can read and write the file, enhancing security.
• Windows: Move the kaggle.json file to C:\Users\<Windows-username>\.kaggle\. Create the .kaggle directory if it doesn't exist.

Your first request

With the Kaggle API client installed and configured, you can now make your first request to verify that everything is set up correctly. A common first request is to download a public dataset, such as the widely used Titanic dataset from a competition.

Downloading a competition dataset

The Kaggle API client provides commands for interacting with various Kaggle resources. To download a competition dataset, you use the kaggle competitions download command.

1. Identify a public competition: The Titanic competition is a good starting point because it is always active and its dataset is small. The competition slug (identifier) is titanic.
2. Execute the download command: Open your terminal or command prompt and run:

   kaggle competitions download -c titanic

3. Verify the download: After the command executes, you should see output indicating the download progress. The dataset files (e.g., train.csv, test.csv, gender_submission.csv) will be downloaded as a ZIP archive into your current working directory. Unzip the archive to access the individual files.

This successful download confirms that your Kaggle account, API key, and client setup are all functioning correctly. You have established programmatic access to Kaggle's resources.

Listing datasets or competitions

To explore other available resources, you can use the list command:

• To list available datasets:

  kaggle datasets list

• To list active competitions:

  kaggle competitions list

These commands will display a table of available resources, which can help you identify datasets or competitions of interest for your next steps.

Common next steps

After successfully making your first request, several common paths lead to further engagement with Kaggle's platform:

• Explore more datasets: Use kaggle datasets list to find new datasets and kaggle datasets download -d <owner>/<dataset-name> to download them. Many public datasets are available for various machine learning tasks Kaggle Datasets.
• Participate in competitions: Browse ongoing competitions on the Kaggle Competitions page. For structured learning, consider 'Getting Started' competitions which often have extensive tutorials.
• Utilize Kaggle Notebooks (Kernels): Kaggle provides an in-browser environment for running Python or R notebooks. These notebooks offer free GPU/TPU access and come pre-configured with popular data science libraries, eliminating local setup complexities. You can create a new notebook directly from a dataset page or competition page Kaggle Notebooks.
• Learn with Kaggle Courses: Kaggle offers a series of free, hands-on courses covering various aspects of data science and machine learning, from Python programming to deep learning. These courses are integrated with the notebook environment Kaggle Learn.
• Contribute datasets and notebooks: Once familiar with the platform, you can share your own datasets or publish notebooks to contribute to the community.
• Integrate with Google Cloud Platform: For more extensive compute or storage needs, Kaggle offers integrations with Google Cloud Platform services. This allows users to leverage GCP's infrastructure for larger-scale projects Google Cloud AI Platform documentation.

Troubleshooting the first call

If your initial API call fails, here are common issues and their resolutions:

• 401 Unauthorized or Authentication failed:
  • Incorrect kaggle.json location: Ensure kaggle.json is in the correct directory (~/.kaggle/ on Linux/macOS, C:\Users\<Windows-username>\.kaggle\ on Windows).
  • Incorrect file permissions: On Linux/macOS, verify the kaggle.json file has restrictive permissions (chmod 600 ~/.kaggle/kaggle.json) to prevent other users from reading it.
  • Expired or revoked API key: Generate a new API token from your Kaggle Account Settings and replace the old kaggle.json file.
  • Missing API key: Confirm you have downloaded the kaggle.json file and placed it in the correct location.
• 403 Forbidden:
  • Competition Rules Acceptance: For competition datasets, you must accept the competition rules on the Kaggle website before you can download its data programmatically. Visit the specific competition page and click 'Join Competition' or 'Accept Rules'.
  • Dataset Access Restrictions: Some datasets may have specific licensing or access requirements. Ensure you meet all prerequisites on the dataset's page.
• Kaggle API client not found:
  • Client not installed: Re-run pip install kaggle.
  • Python environment issues: Ensure your terminal or command prompt is using the Python environment where the Kaggle client was installed. You might need to specify the Python executable (e.g., python3 -m pip install kaggle).
• Network connectivity issues:
  • Verify your internet connection.
  • Check if Kaggle's services are experiencing outages (e.g., via their social media or status page, if available).
• General debugging:
  • Use the -v flag with Kaggle API commands for verbose output, which can provide more detailed error messages (e.g., kaggle competitions download -c titanic -v).
  • Consult the Kaggle API documentation for specific command syntax and error codes.