---

Automated Kaggle Project Setup

This repository contains a Makefile to manage Docker containers and GitHub repositories for a new Kaggle competition project.

Moreover, an optional local instance of MLflow is included.

Necessary Prerequisites

• Docker-engine installed on your machine.
• GitHub CLI installed and configured.
• Kaggle CLI installed and configured.
• Your id_rsa.pub is added for ssh access to the container.
• You know how to use a Makefile

Repository Structure

File Name	Description
config.env	Configuration file for the competition setup
Dockerfile	Dockerfile for the Kaggle container
id_rsa.pub	Public ssh key-file (not included in the repo)
Makefile	Contains all commands to setup competition
mlflow	Necessary files for a local MLflow setup

Configuration

Make sure to configure the config.env file with the necessary environment variables:

Variable Name	Description
IMAGE_NAME	Name of the Docker image
TAG	Tag for the Docker image
DOCKERFILE	Path to the Dockerfile
BUILD_DIR	Directory to build the Docker image
ROOT_FOLDER	Root folder for the project
COMPETITION_NAME	Name of the competition
GITHUB_USER	GitHub username
TEMPLATE_REPO	Template repository name
GITHUB_NO_REPLY_MAIL	No-reply email for GitHub
SHM_SIZE	Shared memory size for Docker
MEMORY_LIMIT	Memory limit for Docker container
SWAP_LIMIT	Swap limit for Docker container

Using the Makefile

Initialize a New Competition

make competition-init

Executes a series of commands to set up everything needed to start working on the defined competition. This includes building the Docker image, initializing the project, running the container, setting up Git, and downloading the competition data.

The Kaggle container will restart and run automatically when you restart your workstation.

Remove Everything Related to a Competition

make competition-remove

Executes a series of commands to remove everything related to the competition. This includes stopping and removing the container, deleting the local and remote project, and cleaning up Docker resources.

Set Up Local MLflow Server

Navigate to ./mlflow and run the following command to set up a local instance of MLflow.

The MLflow instance and its related database and object-storage will restart and run automatically when you restart your workstation.

make mlflow-init

---