Course description:

An overview of regression analysis techniques to build predictive models and evaluate the contribution of a variable to change in output.

Topics covered:

• Simple and multiple linear regression
  • Prediction from a regression model
  • Interpretation of regression coefficients
  • Correlation and the coefficient of determination (R-square)
  • Model assumptions
  • Model validation
     
• Model building and variable selection techniques
  • Parsimonious models
  • Multicollinearity
  • Forward, backwards and stepwise variable selection
     
• Categorical predictors  
  • Creating dummy (indicator) variables
  • Interpretation of coefficients of dummy variables
     
• Non-linear regression
  • Polynomial regression
  • Interactions
     
• Graphical methods for data display
  • Bar charts, histograms, scatterplots, box plots etc. What are their differences?
  • Matching graphics to your data type
  • Comparing groups
     
• Scaling techniques
  • Adjusting for variance inflation
  • Adjusting for skewed data
     
• Residual analysis
  • Identifying outliers
  • Identifying leverage points

Attendees should have an understanding of SPSS equivalent to that taught in the Intro to SPSS course and knowledge of statistics equivalent to the Statistics I course.

Course attendees are required to bring their own laptop installed with a licensed copy of SPSS Base version 10 or above.
Laptops are available for hire at an additional cost.

Note: Class sizes are limited to a maximum of 4 attendees.