Chandan220698 / ML_Project_deployment

ML_Project_deployment

Start Machine Learning Project

Software and Accout Requirement.

1. Github Account
2. Heroku Account
3. VS Code
4. GIT CLI

Steps

1. git status
2. git add
3. git commit -m "message" Git commit create the version and push will push this changes to github
4. git push origin main This origin actually is the repo link
5. git remote -v To check repo link

Virtual environment and requirements.txt

1. Creating conda virtual env conda create -p <env_name> python==3.7 -y -p to create venv in project folder itself
2. Activate venv conda activate <env_name>/ or conda activate <env_name>
3. Creating requirement.txt file pip freeze > requirements.txt
4. To install requirements.txt pip install -r requirements.txt

To setpup CI/CD pipeline in heroku we need following information

1. HEROKU_EMAIL: [e-mail]
2. HEROKU_API_KEY: <API_KEY>
3. HEROKU_APP_NAME: ml-deployment-project-ck

BUILD DOCKER IMAGE

1. docker build -t <image_name>:<tag_name> .

image_name should be in lower case and tag_name generally use 'latest'

2. docker images To list Docker Image and get IMAGE_ID
3. Run Docker Image docker run -p 5000:5000 -e PORT=5000 <IMAGE_ID>
4. docker ps To check running container in docker
5. docker stop <container_id> to stop docker container

Notes

If adding '-e .' then we must have setup.py file in root directory. This will create <custom_pkg_name>-egg.info file for every package which contains "init.py" file. Where "-e ." executed inside "requirements.txt". Its actually lunching the "setup.py" hence the egg.info file will get created for all custom packages.