https://github.com/d-kleine/kneed_visualizations

Visualizations for kneed article on www.towardsdatascience.com
https://github.com/d-kleine/kneed_visualizations

kneedle

Last synced: 8 days ago
JSON representation

Visualizations for kneed article on www.towardsdatascience.com

• Host: GitHub
• URL: https://github.com/d-kleine/kneed_visualizations
• Owner: d-kleine
• Created: 2021-10-05T15:59:54.000Z (over 3 years ago)
• Default Branch: main
• Last Pushed: 2023-05-07T17:54:47.000Z (over 1 year ago)
• Last Synced: 2024-12-06T17:18:11.101Z (about 1 month ago)
• Topics: kneedle
• Language: Jupyter Notebook
• Homepage: https://towardsdatascience.com/detecting-knee-elbow-points-in-a-graph-d13fc517a63c
• Size: 268 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Visualizations for kneed

On the *Towards Data Science* blog , I have published a scientific article that deals with the handling of the Python package "kneed". In this article, the "kneed" underlying algorithm Kneedle is presented. It shows how to identify knee or elbow points (i.e., data points at which the function graph rises or falls the most) with Python in the best possible way. Based on the mathematical definition of curvature for continuous functions, Kneedle detects those beneficial data points showing the best balance inherent tradeoffs — called “knees” (curves that have negative concavity) or sometimes “elbows” (curves that have positive concavity) — in discrete data sets. Finally, useful application examples for the detection of knee or elbow points in Machine Learning are shown.

This repository contains the visualizations made for the kneed article on TDS. Identifying the "knee" or "elbow" point in a curve is a common problem in data science, where one needs to determine the optimal number of clusters or components to use in a model. The visualizations in this repository are designed to help users understand how the kneed algorithm works and how to interpret the results. The article can be read here.