ST 590G -- Computation for Data Analysis

  Third Assignment -- due Thursday, 12 November 2009
  (updated 02 Nov)

  Recall our examination of the Federalist Papers files 
  at the sites
   http://www.foundingfathers.info/federalistpapers/fedxx.htm
  where xx takes values from 01 to 85. (Note there are two versions
  of 70: 'fed70a' and 'fed70b' -- choose one.)
   * or * (note change below)
   http://thomas.loc.gov/home/histdox/fed_xx.html
  where again xx takes values 01 to 85.  (Note that this site has
  the second version of #70, as the file is 'fed_70-2.html')
   * or * (a third site)
   http://www.constitution.org/fed/federaxx.htm
  where xx takes values 01 to 85.
   * or * (a fourth site)
   http://www.let.rug.nl/usa/D/1776-1800/federalist/fedxx.htm
  where xx takes values 01 to 85.  (Same two versions as with the
  'FoundingFathers' site.  Note the country of this host.)

  Find the word frequency of the words used by Mosteller & Wallace
  in 'mwwords.dat' for the observations that are needed for the
  following discriminant analysis. (Note: treat adverbial/adjectival
  or plural forms as the same word: "CONSIDERABLY"="CONSIDERABLE", 
    "INNOVATIONS" = "INNOVATION", "VIGOROUS" = "VIGOR", 
    "MATTERS" = "MATTER", "WORKS" = "WORK".)

  Among historians, the general agreement is that
   John Jay wrote 2,3,4,5, and 64;
   James Madison wrote 10, 14, 37, 38, ..., 48;
   no one is sure about 18, 19, 20, 49, 50, ... 58, 62, 63
   and the others were written by Alexander Hamilton.

  
  Your task is to create two datasets for discriminant analysis.
  1) Create one dataset with word frequencies for the papers with known
  authorship.  Include here all of the (14) Madison papers, and at least
  as many Hamilton papers.  (Use at least 14 of the Hamilton papers;
  use all of them if you want.  You can choose to include the Jay papers
  or just delete them from the analysis.)  Include a variable indicating
  the author.

  2) Create a second dataset with the 15 papers where the authorship 
  is not certain.  Include a variable 'author' with missing values.

  ** If you would like to do your own discriminant analysis, my demo's
  are in 'tornado.sas' in the oct09 directory. **