- The "term and definition" list below names and describes
each of the files related to the textbook analyses. The files can be found
in
`http://www.gsu.edu/~dscbms/dsc8450/dos.zip`or in the file at in the directory`~dscbms/files`on the panther and cheetah unix computers. - Go back to DSc8450 home
`hatco`- The data set containing only the 100 observations. Each SAS analysis program file reads this file automatically.
`hatco.clus (.c)`- A cluster analysis.
`hatco.clus.pca (.cp)`- A cluster analysis with variables preprocessed using principle components analysis.
`hatco.cor`- PROC COR
`hatco.cor.manl (.cm)`- Using IMSL to manually calculate correlations.
`hatco.disc.univ (.dun)`- PROC UNIVARIATE on discriminant analysis data.
`hatco.disc2 (.d2)`- PROC DISCRIM without any holdout sample.
`hatco.discHold.2 (.dh2)`- PROC DISCRIM with separate calibration and holdout samples. This example is somewhat tortured in order to exactly replicate the calibration and holdout samples of Hair et al's. The torture involves arcane extra SAS code in DATA steps to subdivide the data.
`hatco.discHold.2.can1 (.dhc)`- Same as
`hatco.discHold.2`with additional plot of data relative to the discriminant function can1 and with the option ANOVA that provides univariate statistics for testing the hypothesis that the class means are equal in the population for each variable. `hatco.discHold.3 (.dh3)`- This example is NOT covered in this class. A 3 group discriminant analysis is much more complicated than a 2 group analysis.
`hatco.discRndm (.dr)`- PROC DISCRIM is used on a randomly selected pair of calibration (60%) and holdout (40%) samples. The method of selecting the subsamples is extremely crude and produces the desired percentages only approximately. (Another method for randomizing is shown as a SAS example in the link "Introduction to SAS".
`hatco.discU (.du)`- PROC DISCRIM uses Lachenbruch's U method for validation.
`hatco.fac`- PROC FACTOR is applied to only 6 variables as in Hair et al's example.
`hatco.fac7 (.fc7)`- PROC FACTOR is applied to all 7 variables.
`hatco.manov (.man)`- This example is NOT covered in this class. A 3 group discriminant analysis is much more complicated than a 2 group analysis.
`hatco.multiv.normal (.mnl)`- Using IMSL to manually perform a test of the assumption that a set of variables follow a multivariate normal probability distribution.
`hatco.princ.univ (.pru)`- Using PROC PRINCOMP to perform a principle components analysis.
`hatco.reg`- PROC REG
`hatco.reg.resids (.res)`- PROC PLOT residuals plot for which unique plotting symbols are constructed.
`hatco.scatter (.sct)`- PROC PLOT squeezing several plots onto a printed page.
`hatco.standard (.std)`- PROC STANDARD for standardizing variable values.
`hatco.univ (.unv)`- PROC UNIVARIATE
- Go back to DSc8450 home