Using streamlit to set the threshold for a classifier

📓 This code accompanies the blog post at https://www.timvink.nl/streamlit-threshold-app/

Intro

Machine learning classifiers can make binary predictions by setting a threshold on the predicted probabilities. Often it makes sense to set this threshold together with business stakeholders that understand the problem domain.

This repository demonstrates a streamlit app that can facilitate the interactive setting of threshold.

App

Run the app with:

pip install -r requirements.txt
streamlit run app.py

This is what it looks like:

Further reading

Read the full post at timvink.nl/streamlit-threshold-app/

Other resources:

• Streamlit video tutorial Crystal-clear and concise video tutorial by calmcode.io
• Streamlit 101: An in-depth introduction Great example use-case on NYC airbnb data
• Streamlit API reference Overview of all the streamlist elements
• Streamlit community components
• awesome-streamlit list of streamlit resources
• github streamlit topic is a great way to discover more streamlit libraries