.local-bin/docs/kaggle_manager/Usage.md

kaggle_manager

Manages Kaggle datasets and notebooks via the Kaggle API.

Description

This script provides command-line access to Kaggle functionality including:

• Downloading datasets
• Fetching notebooks
• Submitting competition entries

Prerequisites

You must have a kaggle.json file at ~/.kaggle/kaggle.json containing your Kaggle credentials:

{
  "username": "your_username",
  "key": "your_api_key"
}

Usage

./kaggle_manager.sh [options]

Options

• -d, --dataset <name>: Download a dataset (format: username/dataset-name)
• -n, --notebook <name>: Download a notebook (format: username/notebook-name)
• -s, --submit <file>: Submit a notebook file to a competition
• -c, --competition <name>: Specify competition name for submission
• -g, --git: Add and commit changes to Git after operation
• -v, --verbose: Enable verbose output
• -h, --help: Display help message

Examples

# Download a dataset
./kaggle_manager.sh -d "zillow/zecon"

# Download a notebook
./kaggle_manager.sh -n "jfraeys/my-notebook"

# Submit to competition
./kaggle_manager.sh -s submission.csv -c "titanic"

# Download with Git commit
./kaggle_manager.sh -d "zillow/zecon" -g

Directory Structure

• Datasets: Stored in data/raw/
• Notebooks: Stored in notebooks/

Notes

• The script uses curl to interact with the Kaggle API
• Credentials are extracted from kaggle.json using jq
• Git commits use the message: "Added dataset or notebook"