IMAGINE Discussions

Discuss and share topics of interest using ERDAS IMAGINE the world’s leading geospatial data authoring system.
Showing results for 
Search instead for 
Do you mean 
Reply
Highlighted
New Contributor
Posts: 4
Registered: ‎05-29-2019
Accepted Solution

How to quantify feature (variable) importance in IMAGINE using Random Forest algorithm?

Hello,

 

I need to determine the relative contribution (importance) of a set of features (variables) to a classification using the Random Forest Algorithm. (After I have determined the importance of the features I will then perform a classification based on the most important, or best combination of, features.)

 

I don't have programming or Python skills. Is there a tool or function in IMAGINE that can perform this task?

 

Many thanks,

 

Sarah

 

 

Technical Evangelist
Posts: 832
Registered: ‎10-01-2015

Re: How to quantify feature (variable) importance in IMAGINE using Random Forest algorithm?

Hi Sarah,

 

The Classify Using Machine Learning operator has an output port called Training Attribute Importances, which will give you the information you are looking for.

 

ML.PNG

 

Note that if you then decide to cut down the attributes used you will need to re-train (Initialize) the Random Forest based on that subset of attributes.

 

Cheers

 

 

Ian Anderson
Chief Product Owner, Desktop Remote Sensing
Hexagon Geospatial
New Contributor
Posts: 4
Registered: ‎05-29-2019

Re: How to quantify feature (variable) importance in IMAGINE using Random Forest algorithm?

Thank you, Ian, that's very helpful!