#! /bin/bash

# Variables

Filename=$1
SortColumn=$2

#ByCorr="1"
#ByVariance="2"
#ByEntropy="3"
#SortColumn=$ByCorr

Horizon="0"
Subset="-"
Path="/tmp/effort"
rm -r $Path
mkdir $Path
RankingFile="/tmp/attr_iteration.dat"
rm -r $RankingFile
EMstart="1"
EMend="15"

Subset="-"
NewMMRE="99999"

# Create Subsets
#cp data/coc81modeTypeLangType.csv data/subsets/coc81_all.csv

# Preprocess data files
#Filename="data/subsets/coc81_all.csv"
ConvertedFile=$Filename".converted"
SeedFile=$Filename".seed"
#gawk -f convertCSV.awk Method="COC81" Style="precise" $Filename > $ConvertedFile

# Rank Attributes
#echo "Ranking "$Filename "..."

./attr_iteration $Filename COC81 precise $SeedFile coc81:all attrLC $Subset $Path $EMstart $EMend > $RankingFile

# Try Combinations of Effort Multipliers

# Sort by Correlation
#echo "---------------------------------------"
#echo "----------Attribute Rankings:----------"
#echo "---------------------------------------"
#echo "Column, Correlation, Variance, Entropy"
#echo "---------------------------------------"

sort +$SortColumn -n -r $RankingFile > "/tmp/sort_results.dat"
#cat "/tmp/sort_results.dat"

#echo ""
#echo "---------------------------------------"
#echo "--------Attribute Combinations:--------"
#echo "---------------------------------------"
#echo "Mmre, Sd, Pred30, Columns"
#echo "---------------------------------------"

i=$EMstart
#while [ $i -le $EMend ]

while ((i<=EMend))
do
	#Subset+=$(gawk -v x=$i 'NR==x{print $1}' "/tmp/sort_results.dat")
	NewSubset=$Subset$(gawk -v x=$i 'NR==x{print $1}' "/tmp/sort_results.dat")
		
 	# Try Local Calibration using the Subset.
 	#./effort $Filename COC81 precise $SeedFile".pass2" coc81:all simpleLC $Subset $Path
 	#EffortResult=$(./effort $Filename COC81 precise $SeedFile".pass2" coc81:all simpleLC $Subset $Path)
 	EffortResult=$(./effort $Filename COC81 precise $SeedFile".pass2" coc81:all simpleLC $NewSubset $Path)
 #echo $EffortResult
 	OldMMRE=$NewMMRE
 	NewMMRE=$(echo $EffortResult | gawk '{print $1}')
 	#echo "New MMRE is " $NewMMRE
 	
 	# Test new MMRE and handle Subset Selection
 	#if [ $NewMMRE -l $OldMMRE ]
 	#if ((NewMMRE < OldMMRE))
 	Keep=$(gawk -v a=$NewMMRE -v b=$OldMMRE 'BEGIN{if (a<b) {print 1} else {print 0}}')
 	if ((Keep==1))
 	then
 		Subset=$NewSubset
 		#NewMMRE is current
 	else
 		#Keep the old Subset
 		NewMMRE=$OldMMRE
 	fi
 	
	i=$(($i+1))
done
#done | sort -n +0

#echo "-----------Final Selection:------------"
# note: this uses $Subset instead of $NewSubset
./effort $Filename COC81 precise $SeedFile".pass2" coc81:all simpleLC $Subset $Path


# debugging
#echo "-----------Demonstration:------------"
#Subset="-"
#i=$EMstart
#while ((i<=EMend))
#do
	#Subset=$Subset$(gawk -v x=$i 'NR==x{print $1}' "/tmp/sort_results.dat")
# 	Subset=$Subset$i-
#	EffortResult=$(./effort $Filename COC81 precise $SeedFile".pass2" coc81:all simpleLC $Subset $Path)
# 	echo $EffortResult
#	i=$(($i+1))
#done

#echo "-----------Straw Man:------------"
#i=$EMstart
#while ((i<=EMend))
#do
# 	EffortResult=$(./effort $Filename COC81 precise $SeedFile".pass2" coc81:all simpleLC $Subset $Path)
# 	echo $EffortResult
#	i=$(($i+1))
#done