#!/bin/bash
DESTDIR=${DESTDIR=""}
MyHome=${MyHome=${DESTDIR}${HOME}}
# /* vim: set filetype=sh : */ -*- sh -*- 
# Q0: where will the output be stored
Safe=${Safe=${MyHome}/var/weka}

# 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. (27 datasets)
Datums=${Datums="a2b/agaricus-lepiota a2b/anneal 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"}


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

# Q3: what learners will you try? 
Learners=${Learners="j48 bayes nbk nbcd"}

jAode() { wttp $1 $2 $wAode  | gotwant ; }
j48()   { wttp $1 $2 $wJ48   | gotwant ; }
bayes() { wttp $1 $2 $wBayes | gotwant ; } 
nbk()   { wttp $1 $2 $wNbk   | gotwant ; }
oneR()  { wttp $1 $2 $wOneR  | gotwant ; }
nbcd()  { nbcdtree Mode=1 $1 Mode=2 $2 | gotwant; }

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

# Q4 : what pre-processors will you use
Preps=${Preps="cat tbin fayyadIrani"}
disctree() { dtree2 Pass=1 $1 Pass=2 $1; }
tbin()  { tenbins Pass=1 $1 Pass=2 $1 Pass=3 $1; }

# Q5: how many repeats? 
#Repeats=${Repeats=1}
Repeats=${Repeats=7}

# Q6: how many bins? 
#Bins=${Bins=3}
Bins=${Bins=7}

# 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,goal,train,test,repeats,bin,accuracy,pd,pf,precision"
	for prep in $Preps
	do
	  $prep ${Data}/${datum}.arff > data.arff 
          for((repeats=1;repeats<=$Repeats;repeats++))
	  do
	     one=`nthclass Want=1 data.arff` # negative = class 1
	     two=`nthclass Want=2 data.arff` # positive = class 2
	     seed=$RANDOM
      	     for((bin=1;bin<=${Bins};bin++))
      	     do
		somearff Bin=$bin Bins=$Bins Seed=$seed data.arff
		n1=`instances train.arff`
		n2=`instances test.arff`
                for learn in $Learners
                do
         	   b4="$datum,$learn,$prep,$two,$n1,$n2,$repeats,$bin";
		   $learn train.arff test.arff |
                   abcd Prefix="${b4}" False="${one}" True="${two}"
  done; done; done; done; done
}

Log=$$
(main | tee $Safe/xval.$Log ) 2> $Safe/xval.err.$Log 
echo $Safe/xval.$Log