2022-08-14-frontend.md

DIY Frontend for your ML model

• Goal: build a deployable frontend application for a ML model (regression or classification)
• Dates: from 11 August to 18 August
• Where: #project-of-the-week in DataTalks.Club Slack

Technologies

• Scikit-Learn
• Flask
• Docker
• Streamlit

Note: this is a suggested list of technologies, you can chose alternatives instead

Plan

This is a proposed plan only, you don’t have to follow it day-by-day

Day 1

• Come up with a project idea
• Select the dataset for your project
• Create a github repository
• Share your progress in Slack and in social media

Day 2

• Perform data cleaning for your data in a jupyter notebook
• And try out various models for your data in notebook
• Push your changes to github
• Share your progress in Slack and in social media

Resources:

• Data preparation: https://github.com/alexeygrigorev/mlbookcamp-code/tree/master/course-zoomcamp

Day 3

• Convert the notebook code into flask backend and test it using backend only
• Push your changes to github
• Share your progress in Slack and in social media

Resources:

• Model deployment: https://github.com/alexeygrigorev/mlbookcamp-code/tree/master/course-zoomcamp/05-deployment

Day 4

• Start making frontend of the application in streamlit
• Push your changes to github
• Share your progress in Slack and in social media

Resources:

• https://streamlit.io/

Day 5

• Continue making the project
• Dockerize the application
• Deploy the application (optional)
• Push your changes to github
• Share your progress in Slack and in social media
• Start writing article about the project process if you want

Day 6

• Continue writing the article or update the process in the readme
• Push your changes to github
• Share your progress in Slack and in social media
• Feedback session and doubt clearing day

Day 7

• Rest

Going the extra mile

Extra stuff you can add to the application:

• Writing test in pytest for the application.
• Using gradio in place of streamlit or try both.
• Using fastapi in place of flask.

Materials

• https://towardsdatascience.com/create-an-awesome-streamlit-app-deploy-it-with-docker-a3d202a636e8
• https://www.section.io/engineering-education/how-to-deploy-streamlit-app-with-docker/
• Example project: https://www.youtube.com/watch?v=xl0N7tHiwlw
• Gradio + HuggingFace: https://youtu.be/RiCQzBluTxU

FAQ

Will there be any videos?

Not exactly. We’ll have a plan and help you stay on track. We’ll share links and answer questions, and help each other when we’re stuck

Can we use something else, e.g. Gradio instead of streamlit or TensorFlow instead of Scikit-Learn?

Yes! We only propose a plan, but the choice of technology and implementation is up to you

What if I haven’t trained an ML model previously, can I take part in this event too?

Yes. It might be challenging for you to take part, but we can give you an already trained model, and you’ll need to figure out how to build a frontend for it.

Also, check our ML engineering course: https://github.com/alexeygrigorev/mlbookcamp-code/tree/master/course-zoomcamp. You can take this course to learn how to build ML models.

Projects

List of projects from our participants:

• Student Acceptance prediction by Alexey Grigorev
• Fish Detector using YoloV7 by Pathik Ghugare
• Churn Prediction App by Esteban Encina
• (Add your project above this line - create a PR)