Electron microscopy
  
Gini Loss
- Python Automation and Machine Learning for ICs -
- An Online Book -
Python Automation and Machine Learning for ICs                                                           http://www.globalsino.com/ICs/        


Chapter/Index: Introduction | A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | Appendix

=================================================================================

Gini Loss is often used as the criterion for splitting nodes in decision trees. It is calculated by taking the weighted sum of the Gini Impurities of the child nodes after a split. When building a decision tree, the algorithm selects the split that minimizes the Gini Loss.

The Gini loss is given by,

         Upload Files to Webpages ---------------------------------- [3757a]

where,

  • represents the classes in the dataset.
  • is the probability of randomly choosing an element of class .
  • The sum is taken over all classes.

Equation 3757a is used in decision trees and random forests. Assuming there are two children regions R1 and R2, Figure 3757a shows the plot of Gini loss in Equation 3757a.

Cross Entropy

Figure 3757a. Plot of Gini loss in Equation 3757a. The green dot represents L(Rp). They difference beween the green dot and the grey dot is the change of loss. (Code)

============================================

         
         
         
         
         
         
         
         
         
         
         
         
         
         
         
         
         
         

 

 

 

 

 



















































 

 

 

 

 
=================================================================================