README.md

Model Training and Tracking

Setup includes setting up an EC2 instance, RDS database and S3 bucket.

Both experiment tracking and model registry are done in AWS through Jupyter Notebook.

MLflow was used for Model Experimenting and the best nodel was save in an S3 bucket and alos registered with MLflow.

How to run on aws

- Set up an EC2 instance and install appropriate libraries (Conda is sufficient and MLflow)
- Set up RDS database
- Set up S3 bucket
- Set up AWS credentials
- Change AWS_PROFILE variable with yur AWS credential
- Change TRACKING_SERVER_HOST variable to public DNS of the EC2 instance
- Connect to your EC2 instance and Run your Mlflow Server with ---> mlflow server -h 0.0.0.0 -p 5000 --backend-store-uri (database_type)://(username):(password)@(database_address):(port)/database_name --default-artifact-root s3://(S3_bucket_name)