#!/bin/bash
DESTDIR=${DESTDIR="/srv/bronze/dj"}
MyHome=${MyHome=${HOME}} #${HOME}}
# /* vim: set filetype=sh : */ -*- sh -*- 
# Q0: where will the output be stored
Safe=${Safe=${DESTDIR}/var/weka}
BinDir=${BinDir=${MyHome}/bin}
# Q1: where is your data? 
Data=${Data=${MyHome}/var/data/discrete}

# Q2: what data sets will we run?
# A2: only those with binary discrete classes; i.e. (26 datasets)
Datums=${Datums="a2b/audiology a2b/auto-mpg
                 a2b/breast-cancer a2b/breast-cancer-wisconsin
                 c2d/credit-a c2d/diabetes e2i/ecoli e2i/flag e2i/hayes-roth 
                 e2i/heart-c e2i/heart-h e2i/hepatitis e2i/imports-85 
                 e2i/iris j2p/kr-vs-kp j2p/letter j2p/mushroom
                 q2s/segment q2s/sick q2s/splice q2s/soybean t2z/vowel  
                 t2z/wine t2z/wdbc t2z/waveform-5000"}

#Datum=${Datums=" a2b/breast-cancer c2d/diabetes e2i/heart-c e2i/iris
#                 j2p/mushroom  q2s/soybean t2z/vowel t2z/wine"}

# A2': just a couple 
#Datums=" c2d/diabetes a2b/breast-cancer q2s/sonar t2z/vote "

# Q3: what learners will you try? 
#Learners=${Learners="bayes nbk"}
Learners=${Learners="bayes"}
bayes() { wttp $1 $2 $wBayes | gotwant ; } 
#nbk()   { wttp $1 $2 $wNbk   | gotwant ; }

[ -f  "$Functions" ] && . $Functions

# Q4 : what pre-processors will you use
# Preps=${Preps="cat disctree3 fayyadIrani pkid tbin"}
#Preps=${Preps="cat disctreea disctreeb disctreec disctreed fayyadIrani pkid tbin"}
#Preps=${Preps="cat disctreeb2 disctreeb3 disctreeb4 fayyadIrani pkid tbin"} 
#Preps=${Preps="cat disctreea disctreeb disctreec disctreed disctreee disctreef disctreeg disctreeh disctreei disctreej disctreek disctreel disctreem disctreen disctreeo disctreep  fayyadIrani pkid tbin"}
#disctreea() { dtree3 Pass=1 $1 Pass=2 $1; }
Preps=${Preps="disctree2 disctree3 disctree4"}
disctree2()  { dtree2 DynTree=1 Pass=1 $1 Pass=2 $1; }
disctree3() { dtree3 DynTree=1 Pass=1 $1 Pass=2 $1; }
disctree4() { dtree4 DynTree=1 Pass=1 $1 Pass=2 $1; }
#disctreec() { dtree3 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreed() { dtree3 DynTree=1 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreee() { dtree2 MaxSize=15 GrowthMult=2 Pass=1 $1 Pass=2 $1; }
#disctreef() { dtree2 MaxSize=15 GrowthMult=2 DynTree=1 Pass=1 $1 Pass=2 $1; }
#disctreeg() { dtree2 MaxSize=15 GrowthMult=2 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreeh() { dtree2 MaxSize=15 GrowthMult=2 DynTree=1 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreei() { dtree2 IncSize=50 MaxSize=15 GrowthMult=2 Pass=1 $1 Pass=2 $1; }
#disctreej() { dtree2 IncSize=50 MaxSize=15 GrowthMult=2 DynTree=1 Pass=1 $1 Pass=2 $1; }
#disctreek() { dtree2 IncSize=50 MaxSize=15 GrowthMult=2 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreel() { dtree2 IncSize=50 MaxSize=15 GrowthMult=2 DynTree=1 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreem() { dtree2 IncSize=32 MaxSize=15 GrowthMult=2 Pass=1 $1 Pass=2 $1; }
#disctreen() { dtree2 IncSize=32 MaxSize=15 GrowthMult=2 DynTree=1 Pass=1 $1 Pass=2 $1; }
#disctreeo() { dtree2 IncSize=32 MaxSize=15 GrowthMult=2 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
#disctreep() { dtree2 IncSize=32 MaxSize=15 GrowthMult=2 DynTree=1 SizeChoice=1 Pass=1 $1 Pass=2 $1; }
tbin()  { tenbins Pass=1 $1 Pass=2 $1 Pass=3 $1; }
pkid() { wpkid $1;}

# Q5: how many repeats? 
#Repeats=${Repeats=3}
 Repeats=${Repeats=10}

# Q6: how many bins? 
#Bins=${Bins=5}
Bins=${Bins=10}

# All right then. Lets go! 
mkdir -p $Safe                # ensure safe place exists
Tmp=`mktemp -d`               # make a sandbox where only you will play
trap "rm -rf $Tmp" 0 1 2 3 15 # leave nothing behind when you quit 
cd $Tmp                       # go to the sandbox

main() {
  set -x
  for datum in $Datums
  do
  	echo "#data,learner,prep,train,test,repeats,bin,goal,a,b,c,d,accuracy,pd,pf,precision,bal"
	for prep in $Preps
	do
	  $prep ${Data}/${datum}.arff > data.arff 
          for((repeats=1;repeats<=$Repeats;repeats++))
	  do
	     seed=$RANDOM
      	     for((bin=1;bin<=${Bins};bin++))
      	     do
		cat data.arff | someArff --seed $seed --bins $Bins --bin $bin
		goals=`cat data.arff | classes --brief`
		n1=`instances train.arff`
		n2=`instances test.arff`
                for learn in $Learners
                do
		   $learn train.arff test.arff > results.csv
		   cp results.csv $Safe/results.csv
		   for goal in $goals
		   do
		     b4="$datum,$learn,$prep,$n1,$n2,$repeats,$bin,$goal";
		     cat results.csv | ${BinDir}/abcd -g "$goal" -p "${b4}" -d 1 | tail -n 1
                   #abcd Prefix="${b4}" False="${one}" True="${two}"
		   done
done; done; done; done; done
}

Log=$$
Start=`date +%H:%M:%S%t%m/%d/%Y`
(main | tee $Safe/xval.$Log ) 2> $Safe/xval.err.$Log 
cat $Safe/xval.$Log | sort -t, -n -k 17,17 | malign > $Safe/myresults.$Log.csv
#doExtras $Safe/xval.$Log | sort -t, -n -k 19,19 | malign > $Safe/myresults.$Log.csv

echo "Computing acc comparison"
winlosstie --input $Safe/myresults.$Log.csv --fields 17 --perform 13 --key 3 --95 --high > $Safe/crossvalaccmw.$Log.csv
echo "Computing pd comparison"
winlosstie --input $Safe/myresults.$Log.csv --fields 17 --perform 14 --key 3 --95 --high > $Safe/crossvalpdmw.$Log.csv
echo "Computing prec comparison"
winlosstie --input $Safe/myresults.$Log.csv --fields 17 --perform 16 --key 3 --95 --high > $Safe/crossvalprecmw.$Log.csv
echo "Computing bal comparison"
winlosstie --input $Safe/myresults.$Log.csv --fields 17 --perform 17 --key 3 --95 --high > $Safe/crossvalbalmw.$Log.csv
#echo "Computing npf comparison"
#winlosstie --input $Safe/myresults.$Log.csv --fields 19 --perform 18 --key 3 --95 --high > $Safe/crossvalnpfmw.$Log.csv
#echo "Computing fmeas comparison"
#winlosstie --input $Safe/myresults.$Log.csv --fields 19 --perform 19 --key 3 --95 --high > $Safe/crossvalfmeasmw.$Log.csv

End=`date +%H:%M:%S%t%m/%d/%Y`
echo $Start "to" $End 
echo $Start "to" $End >> myresults.$Log.csv
echo $Safe/xval.$Log