MLOps-PSET2

Problem Statement:

Build an ML app with the following features in your code 

• Should have the template from cookie cutter https://drivendata.github.io/cookiecutter-data-science/
• Should have all the python files stored in proper directories  ( modularity )
• Should have DVC pipelines implemented into the work flow 
• Should have implemented the CI pipelines from cml, to compare the accuracies between the two commits
• Should have deployed the above code deployed to either AWS or AZURE

Note : The ML model can be anything from a basic regression process to the advanced classification problem 

Submission : Your github URL of the repo and the URL of your deployed website on AWS or Azure

Marking : The scoring for this assignment will be based on how effective the design methodologies are followed ( for CI and CD ) as mentioned in lectures.

Steps:

Step 1: Create Git repo create DagsHub repo: https://dagshub.com and connect to the github repo

Step 2: install DVC configure dvc:

    dvc init

    dvc remote add origin https://dagshub.com/anibhush/MLOps-PSET2.dvc
    dvc remote modify origin --local auth basic
    dvc remote modify origin --local user anibhush
    dvc remote modify origin --local password <DAGSHUB_TOKEN>

    dvc pull -r origin
    dvc add data/raw
    dvc push -r origin    

Step 3:

install mlflow

# add the following in the python code!
mlflow.set_tracking_uri("https://dagshub.com/anibhush/MLOps-PSET2.mlflow")
tracking_uri = mlflow.get_tracking_uri()
print("Current tracking uri: {}".format(tracking_uri))

export MLFLOW_TRACKING_USERNAME=anibhush
export MLFLOW_TRACKING_PASSWORD=$DAGSHUB_TOKEN