Two outlier detection techniques you should know in

 By Medium - 2021-03-21

Description

An outlier is an unusual data point that differs significantly from other data points. Outlier detection is something tricky that should be done carefully. Elliptic Envelope and IQR are commonly used…

Summary

  • 2021 Elliptic Envelope and IQR-based detection An outlier is an unusual data point that differs significantly from other data points.
  • The intuition behind the Elliptic Envelope (Image by author) The Elliptic Envelope method considers all observations as a whole, not individual features.
  • Apply the elliptic envelope to detect outliers There seems to be a very small proportion of outliers in our dataset.
  • As you know the total of observations, you can get an approximate value for the proportion of outliers.

 

Topics

  1. Backend (0.28)
  2. Security (0.2)
  3. Coding (0.14)

Similar Articles

Datasets should behave like git repositories

By DAGsHub Blog - 2021-01-18

Create, maintain, and contribute to a long-living dataset that will update itself automatically across projects, using git and DVC as versioning systems.