Split a dataset into test and training sets.

* ------------------------------------------------------------------------------;
* SAS Code                                                                      ;
*-------------------------------------------------------------------------------;

data propval;
   infile "c:/datasets/propval.txt" firstobs=15;
   input y x1-x9;

proc surveyselect data=propval method=srs seed=43543 outall
  samprate=0.5 out=subsets;

* Values of Selected variable: 1 means selected for training set, 0 means test set;
proc print data=subsets;
   title "Subsets Dataset";

data training;
   set subsets;
   if Selected=1;
  
proc print;
  
data test;
   set subsets;
   if Selected=0;
   
proc print;

#-------------------------------------------------------------------------------
# R code
#-------------------------------------------------------------------------------

# Read and print data frame.
propval = read.table("c:/datasets/propval.txt", header=T, skip=13)
cat("propval data frame:\n")
print(propval)

# Set seed for random number generator so everyone 
# using this code gets the same results. 
set.seed(34839)

# Select half of the rows to derive model (training),
# select the other half for validation (test).
s = sample(nrow(propval), size=as.integer(nrow(propval) / 2))
print(s)

training = propval[s,]
cat("training data frame:\n")
print(training)

test = propval[-s,]
cat("test data frame:\n")
print(test)