/outliers

Outlier Detection & Removal | Example code and own notes while taking the course "Intro to Machine Learning" on Udacity.

Primary LanguageJupyter Notebook

Outliers - Detection & Removal

Outlier Detection & Removal | Example code and own notes while taking the course "Intro to Machine Learning" on Udacity.

What causes outliers?

  1. Sensor malfunction
  2. Data entry errorrs
  3. Freak event

1 and 2 should be ignored but you have to pay attention the 3rd one!! The 3rd one is like fraud or anomaly detection.

Outlier removal strategy

  1. Train
  2. Remove ~ 10 %
  3. Train again

Observe the training score and start again from 2nd step.