CE576 Highway Design and Traffic Safety - Spring 08

Project 2: Due Date March 31 2008 [15-pages limit]

 Discriminant analysis is used to statistically distinguish between two or more groups of cases by weighting and linearly combining the discriminating variables is some fashion so that the groups are forced to be as statistically distinct as possible. This project is based on the 2003 crash dataset and  involves the use of discriminant analysis to answer the following question:

Based on the data available, can driver 1s be differentiated by age based on characteristics of the crashes in which they were involved? The term, characteristics of the crashes, is meant to include all of the data in the record.

 The following tasks should be part of the activities that you execute to answer the above question:

1. Compare the crash rate for different age groups [less than 20 years, 21-25, 26-30, .., .., more that 90 years]
2. Based on the above comparison, define the age groups used in the study
3. Recode nominal crash characteristics (variables) into binary or interval format
4. Use cross tabulation and correlation statistics, or alternatively use on-way AVOA to examine the relationship between different crash characteristics (variables) and different age groups
5. Based on the cross tabulation/(ANOVA) analysis, choose the appropriate crash characteristics (variables) to be included in the study
6. Examine the correlation (interdependency) among candidate variables
7. Use the correlation tables to determine the variables to be used in the study
8. Discuss important features in discriminant analysis and determine which features you will use in the study [Direct vs. Stepwise, Bayesian Prior Probability, number of functions, etc.]
9. Conduct discriminant analysis using SAS, SPSS, or any other statistical software
10. Present and discuss the output of the discriminant analysis. Examine the within-group correlation and the possible use of sequential models, modifying the age groups, or modifying the prior probability to improve the discriminating power of the resulting functions

Download the data set: text format        SPSS format

Download the List of variables for the working file