© 2019 by the PA CONSULTING GROUP

  • Black Twitter Icon
  • Black LinkedIn Icon

Data Driven Organisation website

AI & Analytics

Feature selection

Aspiring data scientists with a wish to deepdive on the most important technologies

Aimed at:

Delivery method:

Interactive classroom training combining theory with practical exercises

Prerequisites:

Required preparation:

Knowledge of programming basics (preferably, but not necessarily, in R)

Bring a laptop with RStudio installed

2-3 hours

Foundation

Duration:

Skill level:

Prefered group size:

8 participants per trainer (scalable)

Course description

So you've collected at lot of data, and found or built a large number of features to explain the phenomenon of interest. How do you make sure your models do not become too big and unwieldy, while maintaining good performance? Feature selection comes to the rescue. We will talk about filter, wrapper, and embedded methods of feature selection, and apply them in a classification problem using a Random Forest algorithm.

Learning objective

Upon completion of this training, participants will be able to understand the powers and weakness of different variables working together; and how to manage the features in order to increase performance