uscrime {nipals}

U.S. Crime rates per 100,00 people

Description

U.S. Crime rates per 100,00 people for 7 categories in each of the 50 U.S. states in 1977.

Usage

uscrime

Format

A data frame with 50 observations on the following 8 variables.

state

U.S. state

murder

murders

rape

rapes

robbery

robbery

assault

assault

burglary

burglary

larceny

larceny

autotheft

automobile thefts

Details

There are two missing values.

Source

Documentation Example 3 for PROC HPPRINCOMP. http://documentation.sas.com/api/docsets/stathpug/14.2/content/stathpug_code_hppriex3.htm?locale=en

References

SAS/STAT User's Guide: High-Performance Procedures. The HPPRINCOMP Procedure. http://support.sas.com/documentation/cdl/en/stathpug/67524/HTML/default/viewer.htm#stathpug_hpprincomp_toc.htm

Examples

library(nipals)
head(uscrime)

# SAS deletes rows with missing values
dat <- uscrime[complete.cases(uscrime), ]
dat <- as.matrix(dat[ , -1])
m1 <- nipals(dat) # complete-data method

# Traditional NIPALS with missing data  
dat <- uscrime
dat <- as.matrix(dat[ , -1])
m2 <- nipals(dat, gramschmidt=FALSE) # missing 
round(crossprod(m2$loadings),3) # Prin Comps not quite orthogonal
  
# Gram-Schmidt corrected NIPALS
m3 <- nipals(dat, gramschmidt=TRUE) # TRUE is default
round(crossprod(m3$loadings),3) # Prin Comps are orthogonal