#! /bin/bash
# This generates 10 experts from the training data

MyRoot="../../../trunk/"
cd ..

if [ ! -d /tmp/$USER ]
then
mkdir /tmp/$USER
fi

# Set up data to demonstrate usage
#Data="/tmp/$USER/nasa93_all.csv.numeric"
Data="/tmp/$USER/coc81_all.csv.numeric"
RawData=$MyRoot"data/coc81.csv"
Train="/tmp/$USER/train.csv"
Test="/tmp/$USER/test.csv"
echo -n "" > $Train
echo -n "" > $Test
#echo "Preprocessing data file"
gawk -f $MyRoot"baker_lc/convertcsv.awk" Method="COC81" Style="precise" $RawData > $Data

# Print a header line of the attribute names.
# IMPORTANT - This is only for COC81 attributes and would need updated for COCII
head -1 $RawData |
gawk 'BEGIN{FS=OFS=","}
{
  for(i=1;i<=(NF-2);i++)
  {
    if (index("rely,data,cplx,time,stor,virt,turn,acap,aexp,pcap,vexp,lexp,modp,tool,sced", $i)!=0)
    {
      if(i<(NF-2)){printf("%s,",$i)}else{printf("%s\n",$i)}
    }
  }
}
'

#echo "Preprocessed data stored at "$Data
N=$(gawk 'END{print NR}' $Data)
i=$(gawk -v X=$RANDOM -v N=$N 'BEGIN{print int(N*(X/32767))}')
gawk -v X=$i -v Train=$Train -v Test=$Test '{if(NR==X){print $0 >> Test}else{print $0 >> Train}}' $Data
#echo "Training file generated and stored at "$Train
#echo "Test file generated and stored at "$Test
#echo "This may take a few minutes because COCOMOST has to run N+1 times and there are a lot of I/O operations to make the training files."
#echo ""

# Run range
#./range $Train $Test
Seed=$RANDOM
#echo "Seed="$Seed

#Arguments: TrainingFile, Seed, NumberOfExperts, TestSizeOfExperts

./cocomostExperts $Train $Seed 10 0.33