All public courses using SAS are held at SAS, Marlow:

and cost for two-day courses 1000 GBP (plus VAT) per person which includes lunch and refreshments. Reduction to 900 GBP (plus VAT) per person for multiple bookings from the same organisation.

To book a place on a course either click here, or phone +44 (0)118 934 4265, or print and complete an application form (pdf format 115 Kb) .

Course dates for 2008/2009 :

• Propensity Scoring Models: 13th-14th Nov 2008, 12th-13th Feb 2009, 11th-12th June 2009, 5th-6th Nov 2009
  
• Segmentation of a Customer Database: 3rd-4th Nov 2008, 26th-27th March 2009, 24th-25th Sept 2009
  
• Planning & Interpreting of Marketing Campaigns: 18th-19th Sept 2008, 4th-5th June 2009
  

Propensity Scoring Models

Using the SAS^® System

The Course

Who will buy a product? Who will repay a loan? Who will display any particular behaviour with a commercial value? This course will take you through the whole process using The SAS^® System. We look at how to build and assess the model, how to choose and modify predictors, how to score (and re-score) individuals in a large database, how to validate and maintain your models, how to deal with the biases that can arise through repeated use of the models, how to exploit fully the opportunities (and challenges) that arise when data is available on very large numbers of customers. Although the main attention will be on Logistic Regression (using PROC LOGISTIC) we will also consider the role that might be played by Principal Components for smoothing the data prior to building the models and other data smoothing techniques.

We will present all the tools that you need during the first day of the course. We will begin by taking a rapid look at the development of propensity scoring in order to bring out how it has changed from its origins and to examine how the demands made on the technique have changed, as have the criteria for measuring is success. We cover briefly the theory of logistic regression and discuss the importance of the concepts of Odds and the Odds ratio.

From there, we help you to explore the types of data structure that will be likely to arise in a commercial context, discuss the advantages and difficulties that may arise from different structures and look at the extent to which simple correlations might be helpful (when your predictors are continuous measurements) or Chi-square analysis (when they are nominal or banded variables).

We will show you why it is never necessary to use dummy (indicator) variables, though it might sometimes be helpful to do so, since it may make it easier to use some form of automatic selection of predictor variables; we will look at different ways of pre-processing the possible predictor variables to encourage them to perform better (Principal Components, percentile splits - and other transformations).

We will discuss the important features of the main SAS procedure for carrying out logistic regression (PROC LOGISTIC) and will demonstrate that sometimes it can be advantageous to use simpler modelling procedures like PROC REG or CHAID analysis. We will discuss the relevance of neural nets in this context.

During the course there will be several short exercises where you will practice the techniques which have been discussed. There is also an ongoing workshop throughout the course where you can work on data of realistic structure and quantity to solve an extended problem requiring the development of scoring models. There will also be plenty of opportunity for discussion where you will have the opportunity to present any special problems you may have, or have discovered during the workshop, and exploit the experience of the trainers and the other participants.