Logistic Regression Modelling (Credit Scoring) using SAS -step by step

The course promises to explain concepts in a crystal clear manner. It goes through the practical issue faced by analyst.

Course Package:

• 72 mp4 HD Videos
• 4 Excel File
• 2 Word Document
• 8 PDF Files

Some of the discussion items would be:

• How to clarify objective and ensure data sufficiency.
• How to go for variable selection? How to deal with numeric variables and character variables.
• What is the approach to take when you starting with 1000 variables vs. when you are starting with 150 variables.
• Understanding multi collinearity removal steps and recommendations.
• Understanding step wise regression
• Understanding different section of logistic regression output
• Understand model validation and coefficient stability check
• Understanding the strength of the model - KS, Gini

Course Outline:

Section 1 – Introduction to Model

• Understand what is a model? What is a credit score? What is modelling?
• Where it is used?
• What benefit it brings?
• Discuss various kinds of score / models.
• What makes a typical scorecard

Section 2 – Data design

• Steps of model building
• Understand, terms associated with modelling like historical window, performance period etc.
• Understand, how to select performance window
• Understand about the characteristics, which can be considered, for model building.
• Practical scenarios associated with model design discussion

Section 3 - Data Audit and Treatment

• How to build data for model building
• How to apply exclusions before getting into the data
• Learn data investigation techniques through simple but powerful techniques. The trick lies in getting into the more details and finer prints of the output.
  
  
  
  • Proc contents
  • Proc print
  • Proc means
  • Proc freq
• Examining bi-variate plot

Section 4 – Variable selection

• How to go for variable selection
• Different variable selection techniques such as
  
  
  
  • Chi Square
  • Stepwise Regression
  • Info value
  • Fishers linear discriminant function
  • Cramer's V, Phi Square
• Guidance about, which technique to apply when.

Section 5 – Multi collinearity Treatment

• What is multi collinearity?
• Recommendations about how remove it?
• Understand output of each steps of multi collinearity removal.
• Understand terms associated with the same
  
  
  
  • VIF
  • Factor loading
  • Wald chi square (bi variate and multi-variate)

Section 6 – Iterate for final model development

• How to apply forward / backward stepwise regression
• How to decide about final number of variables in the model
• Deep dive in logistic regression output
• Understanding terms associated with logistic regression output
  
  
  
  • Log likelihood ratio
  • AIC
  • SC
  • Concordance
  • Somer's D

Section 7 – Validate model

• KS
• Rank ordering
• Gini Statistics
• Coefficient stability
• Scoring
• Model usage guideline
• Model presentation guideline

Keywords:

logistic, logistic regression, logistic regression using SAS, credit scoring, SAS, scoring, modelling, data mining, video course