#! /bin/bash

# This version assumes the attributes have already been ranked and saved to $RankingFile

# Arguments
# ./attr $Filename $Path $Ranker $TestFile $TrainFile $LogFile $Horizon $RankingFile $SortingFile $Eval $Search

Filename=$1
Path=$2
Ranker=$3
TestFile=$4
TrainFile=$5
LogFile=$6
Horizon=$7
RankingFile=$8
SortingFile=$9
Eval=${10}
Search=${11}

EMstart=1
EMend=15
Stale=$Horizon
Subset=""
i=$EMstart

# Set up the Ranking criteria for evaluating attributes
if [ $Ranker = "correlation" ]
then
    SortColumn=2
    Reverse=1
fi
if [ $Ranker = "variance_high" ]
then
    SortColumn=3
    Reverse=1
fi
if [ $Ranker = "variance_low" ]
then
    SortColumn=3
    Reverse=-1
fi
if [ $Ranker = "entropy_high" ]
then
    SortColumn=4
    Reverse=1
fi
if [ $Ranker = "entropy_low" ]
then
    SortColumn=4
    Reverse=-1
fi

# Set up the Evaluation criteria for finding improvement
if [ $Eval = "mmre" ]
then
    EvalColumn=1
    EvalDef=1
fi
if [ $Eval = "sd_mre" ]
then
    EvalColumn=2
    EvalDef=1
fi
if [ $Eval = "pred30" ]
then
    EvalColumn=3
    EvalDef=0
fi
if [ $Eval = "correlation" ]
then
    EvalColumn=4
    EvalDef=0
fi

if (($EvalDef==1))
then
    NewValue=99999
else
    NewValue=-99999
fi

if [ $Search = "backward" ]
then
	Reverse=$((Reverse*-1))
fi

if ((Reverse==1))
then
    sort -t, -nrk$SortColumn $RankingFile > $SortingFile
else
    sort -t, -nk$SortColumn $RankingFile > $SortingFile
fi

while ((i<=EMend))
do
	NewSubset=$Subset$(gawk -v x=$i 'BEGIN{FS=","} NR==x{print $1}' $SortingFile)
	
	# If this is a backward elimination search then invert the subset.
	if [ $Search = "backward" ]
	then
	    tmpSubset=$(./inverse $NewSubset $EMstart $EMend)
	else
	    tmpSubset=$NewSubset
	fi
 	# Try Local Calibration using the Subset.
 	#EffortResult=$(./effort $Filename $Path "simpleLC" $TestFile $TrainFile $tmpSubset $LogFile"."$i)
	EffortResult=$(./effort $Filename $Path "simpleLC" $TrainFile $TrainFile $tmpSubset $LogFile"."$i)
 		
 	OldValue=$NewValue
 	NewValue=$(echo $EffortResult | gawk -v EvalColumn=$EvalColumn 'BEGIN{FS=","} {print $EvalColumn}')
 	
 	# Test the new value and handle Subset Selection
 	Keep=$(gawk -v a=$NewValue -v b=$OldValue 'BEGIN{print a<b}')
 	if (($Keep==$EvalDef))
 	then
	    # Improvement!
	    Subset=$NewSubset
	    Stale=$Horizon
	    #NewValue is current already
 	else
	    #Keep the old Subset by not updating it
	    Stale=$(($Stale-1))
	    NewValue=$OldValue
 	fi
 	
	# If things have gotten stale it's time to move on, otherwise continue.
	if (($Stale<1))
	then
	    i=999
	else
	    i=$(($i+1))
	fi
done

# Recreate the Final Selection
if [ $Search = "backward" ]
then
    tmpSubset=$(./inverse $Subset $EMstart $EMend)
else
    tmpSubset=$Subset
fi
./effort $Filename $Path "BasicLC" $TestFile $TrainFile $tmpSubset $LogFile".final"