Variable Selection and Enterprise Miner

1
Section 4.1

• Variable Selection and Enterprise Miner

2
Objectives

• Discuss the need for variable selection.
• Explain the methods of variable selection
  available in Enterprise Miner.
• Demonstrate the use of different variable
  selection methods.

3
The Curse of Dimensionality
1D
2D
3D
4
Methods of Variable Selection

• Stepwise Regression
• Decision Trees
• Variable Selection Node

5
Stepwise Regression

• Uses multiple regression p-values to eliminate
  variables.
• May not perform well with many potential input
  variables.

6
Decision Trees

• Grow a large tree.
• Retain only the variables important in growing
  the tree for further modeling.

7
Variable Selection Node

• Selection based on one of two criteria
• R-square
• chi-square for binary targets only.

8
R-square variable selection

• EM computes squared correlation for each
  variable rejects those with squared correlation
  less than .005.
• EM evaluates remaining variables using forward
  stepwise R-square regression. Variables that give
  an improvement of less than .0005 are rejected.
• For binary targets EM performs a logistic
  regression with forward stepwise selection.

9
Chi-square selection

• Variable selection is performed using binary
  splits for maximizing the chi-square value of a
  2x2 frequency table.

10
Conclusion

• Variable selection node automates many strategies
  for variable selection

11
Intro to Neural Networks
12
Biological Inspiration for an Artificial Neural
Network (ANN)
13
Resulting ANN Based on Hebbian Learning
14
Application of ANN in Banking
15

• Build an ANN for the Churn Model
• Discuss over fitting
• Discuss over training