*********** txns.sh ************

#!/bin/bash
#SBATCH --job-name=txns
#SBATCH --output=../analysis/transactions_table.txt
#SBATCH --error=logs/txns_%j.err
#SBATCH --ntasks=1
#SBATCH --mem=10Gb
#SBATCH --time=4:00:00
#SBATCH --partition=netsi_standard

CASHDEP=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "cash-dep" | wc -l)
CASHDEP_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "cash-dep" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

BANKDEP=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "bank-dep" | wc -l)
BANKDEP_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "bank-dep" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

P2P=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "p2p" | wc -l)
P2P_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "p2p" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

CASHPAY=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "cash-pay" | wc -l)
CASHPAY_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "cash-pay" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

BILLPAY=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "bill-pay" | wc -l)
BILLPAY_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "bill-pay" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

MINSPAY=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "mins-pay" | wc -l)
MINSPAY_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "mins-pay" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

CASHWTD=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "cash-wtd" | wc -l)
CASHWTD_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "cash-wtd" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

BANKWTD=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "bank-wtd" | wc -l)
BANKWTD_a=$(grep -v -F "2016-04-" ../data/transactions.csv | grep -F "bank-wtd" | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

TOTAL=$(grep -v -F "2016-04-" ../data/transactions.csv | wc -l)
TOTAL_a=$(grep -v -F "2016-04-" ../data/transactions.csv | cut -d, -f6,6 | awk '{ SUM += $1} END { printf "%d",  SUM }')

echo "cash-dep bank-dep p2p cash-pay bill-pay mins-pay cash-wtd bank-wtd total"
echo $CASHDEP $BANKDEP $P2P $CASHPAY $BILLPAY $MINSPAY $CASHWTD $BANKWTD $TOTAL
echo $CASHDEP_a $BANKDEP_a $P2P_a $CASHPAY_a $BILLPAY_a $MINSPAY_a $CASHWTD_a $BANKWTD_a $TOTAL_a




************ follow.sh ************

#!/bin/bash
#SBATCH --job-name=follow
#SBATCH --output=logs/follow_%j.out
#SBATCH --error=logs/follow_%j.err
#SBATCH --ntasks=1
#SBATCH --mem=100Gb
#SBATCH --time=24:00:00
#SBATCH --partition=netsi_standard

module load python/3.7.1

python follow-the-money_0.2.0/follow_the_money.py ../data/transactions.csv config.json ../data --lifo --smallest 1
(head -1 ../data/flows_lifo.csv && tail -n +2 ../data/flows_lifo.csv | sort -S 50G -T /scratch/mattsson.c/tmp -t, -k1 -s) > ../data/flows_lifo_time.csv




************ flows.sh ************

#!/bin/bash
#SBATCH --job-name=flows
#SBATCH --output=../analysis/flows_table.txt
#SBATCH --error=logs/flows_%j.err
#SBATCH --ntasks=1
#SBATCH --mem=10Gb
#SBATCH --time=4:00:00
#SBATCH --partition=netsi_standard

TOTAL=$(grep -F "2016-" ../data/flows_lifo.csv | wc -l)
TOTAL_a=$(grep -F "2016-" ../data/flows_lifo.csv | cut -d, -f2,2 | awk '{ SUM += $1} END { printf "%d",  SUM }')

echo "total"
echo $TOTAL
echo $TOTAL_a




************ analyse.sh ************ 

#!/bin/bash
#SBATCH --job-name=analyze
#SBATCH --output=logs/analyze_%j.out
#SBATCH --error=logs/analyze_%j.err
#SBATCH --ntasks=1
#SBATCH --mem=100Gb
#SBATCH --time=24:00:00
#SBATCH --partition=netsi_standard

module load python/3.7.1

# motifs
python follow-the-money_0.2.0/analysis/motifs.py ../data/flows_lifo.csv ../analysis/ --prefix c3_ --circulate 3 --timewindow '(2016-05-01 00:00:00,2016-10-01 00:00:00)'
python follow-the-money_0.2.0/analysis/motifs.py ../data/flows_lifo.csv ../analysis/ --prefix c4_ --circulate 4 --timewindow '(2016-05-01 00:00:00,2016-10-01 00:00:00)'

# savings
python follow-the-money_0.2.0/analysis/users_savings.py ../data/flows_lifo.csv ../analysis/ --prefix 30_ --timewindow '(2016-05-01 00:00:00,2016-10-01 00:00:00)' --max_days 30