#! /bin/bash

MyRoot="../../../../trunk/"
Path="tmp/Boosting3"
rm -r $Path
DataPath=$Path"/data"
mkdir -p $DataPath

Method="COC81"
Style="precise"

# output the first line of the csv output describing the field names
echo "#FILENAME,I,LEARNER,ESTIMATE,ACTUAL,MRE"
#Seed=$RANDOM
#echo "#SEED="$Seed

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


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

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

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

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


# try experiment on full datasets
Train=$DataPath"/coc81_all.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_all.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

# try experiment on coc81 data subsets
Train=$DataPath"/coc81_kind_min.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/coc81_lang_ftn.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/coc81_mode_e.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/coc81_kind_max.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/coc81_mode_org.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/coc81_lang_mol.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base


# try experiment on nasa93 data subsets
Train=$DataPath"/nasa93_project_gro.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_cat2_missionplanning.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_cat2_avionicsmonitoring.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_mode_semidetached.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_project_sts.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_forg_g.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_center_5.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_year_1975.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_year_1980.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_mode_embedded.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base

Train=$DataPath"/nasa93_center_2.csv.numeric"
Base=$(basename $Train)
boosting_oversample_3 $Train $Base