End to End machine learning workflow

Used Car Prediction

Instructions

This repo contains sample code for the End to End machine learning for used car price prediction. Data set used for the project is Vehicle dataset from kaggle

• Suggested to try the project by creating a virtual environment

  python -m venv env_end2endML
  

• Activate the virutal environment
• Install the dependencies using the command

  pip install -r requirements.txt 
  

To train the model :

1. Download the data
2. Modify the configuration according to needs in config.yaml
3. Train the model by running python src/train.py

---

Experiments and artifacts(Code, data, model) tracking using weights and biases

4. After step 3, you will be prompted to enter wandb login details to track the experiments, metrics and artifcats
5. For hyper parameter sweep run wandb agent username/project-name/sweepcode
6. Check previous experiments and sweeps here
7. Check model artifacts with versioning here

---

CI/CD using github actions, unit testing with pytest

This repo contains two workflows

8. Python applicaton work flow to check if training pipeline is not broken after every push/pull.
9. Extend test cases if needed here or here
10. Docker build workflow to check if docker image building is successful after every push/pull.

---

Deploying the model using an webapp

11. Make predictions on the validation set by running python src/predict.py
12. To run the streamlit app streamlit run src/app.py
13. After step 12 open the browser if you are running in your local machine or check step 14
14. Check the public app

Serving model using FAST API

15. To serve using fast API using following commands

cd fastapi 

uvicorn main:app -reload

16. check on local host

Building Docker Image

16. Configure the Docker image by editing the Dockerfile
17. Build the docker image using sudo docker run -p 8051:8051 appname:tag

Followed by container orchestration and model monitoring

This benchmark took approximately 0.5 hours to execute on a Windows 10 laptop with 12GB of RAM and 8 cores at 2GHz.