#! /bin/bash

Method="COC81"
Style="precise"

DoStratify=1
DoNFold=1

Path="~/var/fssExperiment"
rm -r $Path
mkdir -p $Path"/log"
mkdir -p $Path"/data/test"
mkdir -p $Path"/data/train"
DataPath=$Path"/data"
mkdir -p $Path"/ranking/sorted"

# output the first line of the csv output describing the field names
#echo "#FILENAME,I,LEARNER,SEARCH,HORIZON,RANKER,EVAL,MMRE,SD(MRE),PRED30,CORR,COLUMNS"
#echo "#FILENAME,I,SEED,LEARNER,SUBSET,ESTIMATE,ACTUAL,MRE"
echo "#FILENAME,I,TIME,LEARNER,SUBSET,ESTIMATE,ACTUAL,MRE"

#FILENAME,I,LEARNER,SEARCH,HORIZON,RANKER,EVAL,MMRE,SD(MRE),PRED30,CORR,COLUMNS,TIME

# Preprocess the full coc81 and nasa93 datasets
#cp data/coc81modetypelangtype.csv $DataPath"/coc81_all.csv"
cp data/coc81modetypelangtype.csv $DataPath"/coc81_all.csv"
gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_all.csv" > $DataPath"/coc81_all.csv.numeric"
cp data/nasa93.csv $DataPath"/nasa93_all.csv"
gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_all.csv" > $DataPath"/nasa93_all.csv.numeric"

if (($DoStratify==1))
then
	#Make the subsets of coc81 data
	gawk -f csvsubset.awk Col="kind" Want="min" data/coc81modetypelangtype.csv > $DataPath"/coc81_kind_min.csv"
	gawk -f csvsubset.awk Col="lang" Want="ftn" data/coc81modetypelangtype.csv > $DataPath"/coc81_lang_ftn.csv"
	gawk -f csvsubset.awk Col="mode" Want="e" data/coc81modetypelangtype.csv > $DataPath"/coc81_mode_e.csv"
	gawk -f csvsubset.awk Col="kind" Want="max" data/coc81modetypelangtype.csv > $DataPath"/coc81_kind_max.csv"
	gawk -f csvsubset.awk Col="mode" Want="org" data/coc81modetypelangtype.csv > $DataPath"/coc81_mode_org.csv"
	gawk -f csvsubset.awk Col="lang" Want="mol" data/coc81modetypelangtype.csv > $DataPath"/coc81_lang_mol.csv"

	#Convert the subsets of coc81 data to numbers
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_kind_min.csv" > $DataPath"/coc81_kind_min.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_lang_ftn.csv" > $DataPath"/coc81_lang_ftn.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_mode_e.csv" > $DataPath"/coc81_mode_e.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_kind_max.csv" > $DataPath"/coc81_kind_max.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_mode_org.csv" > $DataPath"/coc81_mode_org.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/coc81_lang_mol.csv" > $DataPath"/coc81_lang_mol.csv.numeric"

	#Make the subsets of nasa93 data
	gawk -f csvsubset.awk Col="projectname" Want="gro" data/nasa93.csv > $DataPath"/nasa93_project_gro.csv"
	gawk -f csvsubset.awk Col="cat2" Want="missionplanning" data/nasa93.csv > $DataPath"/nasa93_cat2_missionplanning.csv"
	gawk -f csvsubset.awk Col="cat2" Want="avionicsmonitoring" data/nasa93.csv > $DataPath"/nasa93_cat2_avionicsmonitoring.csv"
	gawk -f csvsubset.awk Col="mode" Want="semidetached" data/nasa93.csv > $DataPath"/nasa93_mode_semidetached.csv"
	gawk -f csvsubset.awk Col="projectname" Want="sts" data/nasa93.csv > $DataPath"/nasa93_project_sts.csv"
	gawk -f csvsubset.awk Col="forg" Want="g" data/nasa93.csv > $DataPath"/nasa93_forg_g.csv"
	gawk -f csvsubset.awk Col="center" Want="5" data/nasa93.csv > $DataPath"/nasa93_center_5.csv"
	gawk -f csvsubset.awk Col="year" Want="1975" data/nasa93.csv > $DataPath"/nasa93_year_1975.csv"
	gawk -f csvsubset.awk Col="year" Want="1980" data/nasa93.csv > $DataPath"/nasa93_year_1980.csv"
	gawk -f csvsubset.awk Col="mode" Want="embedded" data/nasa93.csv > $DataPath"/nasa93_mode_embedded.csv"
	gawk -f csvsubset.awk Col="center" Want="2" data/nasa93.csv > $DataPath"/nasa93_center_2.csv"

	#Convert the subsets of nasa93 data to numbers
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_project_gro.csv" > $DataPath"/nasa93_project_gro.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_cat2_missionplanning.csv" > $DataPath"/nasa93_cat2_missionplanning.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_cat2_avionicsmonitoring.csv" > $DataPath"/nasa93_cat2_avionicsmonitoring.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_mode_semidetached.csv" > $DataPath"/nasa93_mode_semidetached.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_project_sts.csv" > $DataPath"/nasa93_project_sts.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_forg_g.csv" > $DataPath"/nasa93_forg_g.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_center_5.csv" > $DataPath"/nasa93_center_5.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_year_1975.csv" > $DataPath"/nasa93_year_1975.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_year_1980.csv" > $DataPath"/nasa93_year_1980.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_mode_embedded.csv" > $DataPath"/nasa93_mode_embedded.csv.numeric"
	gawk -f convertcsv.awk Method=$Method Style=$Style $DataPath"/nasa93_center_2.csv" > $DataPath"/nasa93_center_2.csv.numeric"
fi


if (($DoNFold==1))
then
  # try nfold on full datasets
  ./nfold $Path "coc81_all.csv"
  ./nfold $Path "nasa93_all.csv"

  if (($DoStratify==1))
  then
	# nfold on coc81 data subsets
	./nfold $Path "coc81_kind_min.csv"
	./nfold $Path "coc81_lang_ftn.csv"
	./nfold $Path "coc81_mode_e.csv"
	./nfold $Path "coc81_kind_max.csv"
	./nfold $Path "coc81_mode_org.csv"
	./nfold $Path "coc81_lang_mol.csv"

	# nfold on nasa93 data subsets
	./nfold $Path "nasa93_project_gro.csv"
	./nfold $Path "nasa93_cat2_missionplanning.csv"
	./nfold $Path "nasa93_cat2_avionicsmonitoring.csv"
	./nfold $Path "nasa93_mode_semidetached.csv"
	./nfold $Path "nasa93_project_sts.csv"
	./nfold $Path "nasa93_forg_g.csv"
	./nfold $Path "nasa93_center_5.csv"
	./nfold $Path "nasa93_year_1975.csv"
	./nfold $Path "nasa93_year_1980.csv"
	./nfold $Path "nasa93_mode_embedded.csv"
	./nfold $Path "nasa93_center_2.csv"
  fi
  exit
fi

# try learners on full datasets
./learners $Path "coc81_all.csv"
./learners $Path "nasa93_all.csv"

if (($DoStratify==1))
then
	# learners on coc81 data subsets
	./learners $Path "coc81_kind_min.csv"
	./learners $Path "coc81_lang_ftn.csv"
	./learners $Path "coc81_mode_e.csv"
	./learners $Path "coc81_kind_max.csv"
	./learners $Path "coc81_mode_org.csv"
	./learners $Path "coc81_lang_mol.csv"

	# learners on nasa93 data subsets
	./learners $Path "nasa93_project_gro.csv"
	./learners $Path "nasa93_cat2_missionplanning.csv"
	./learners $Path "nasa93_cat2_avionicsmonitoring.csv"
	./learners $Path "nasa93_mode_semidetached.csv"
	./learners $Path "nasa93_project_sts.csv"
	./learners $Path "nasa93_forg_g.csv"
	./learners $Path "nasa93_center_5.csv"
	./learners $Path "nasa93_year_1975.csv"
	./learners $Path "nasa93_year_1980.csv"
	./learners $Path "nasa93_mode_embedded.csv"
	./learners $Path "nasa93_center_2.csv"
fi