Add pash bus analytics benchmark

report/benchmarks/bus-analytics/1.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+# Vehicles on the road per day
+
+cat $INPUT_FILE | sed 's/T..:..:..//' | cut -d ',' -f 1,3 | sort -u | cut -d ',' -f 1 | sort | uniq -c | awk -v OFS="\t" "{print \$2,\$1}"

report/benchmarks/bus-analytics/2.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+# Days a vehicle is on the road
+
+cat $INPUT_FILE |
+  sed 's/T..:..:..//' |
+  cut -d ',' -f 3,1 |
+  sort -u |
+  cut -d ',' -f 2 |
+  sort |
+  uniq -c |
+  sort -k1n |
+  awk -v OFS="\t" "{print \$2,\$1}"

report/benchmarks/bus-analytics/3.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+# Hours each vehicle is on the road
+
+cat $INPUT_FILE |
+  sed 's/T\(..\):..:../,\1/' |
+  cut -d ',' -f 1,2,4 |
+  sort -u |
+  cut -d ',' -f 3 |
+  sort |
+  uniq -c |
+  sort -k1n |
+  awk -v OFS="\t" "{print \$2,\$1}"

report/benchmarks/bus-analytics/4.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+# Hours monitored each day
+
+cat $INPUT_FILE |
+  sed 's/T\(..\):..:../,\1/' |
+  cut -d ',' -f 1,2 |
+  sort -u |
+  cut -d ',' -f 1 |
+  sort |
+  uniq -c |
+  awk -v OFS="\t" "{print \$2,\$1}"

report/benchmarks/bus-analytics/5.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+# Hours each bus is active each day
+
+# Records are day, hour, line, bus
+<$INPUT_FILE sed 's/T\(..\):..:../,\1/' | awk -F, '
+!seen[$1 $2 $4] { seen[$1 $2 $4] = 1; hours[$1 $4]++; bus[$4] = 1; day[$1] = 1; }
+END {
+   PROCINPUT_FILEFO["sorted_in"] = "@ind_str_asc"
+   for (d in day)
+     printf("\t%s", d);
+   printf("\n");
+   for (b in bus) {
+     printf("%s", b);
+     for (d in day)
+       printf("\t%s", hours[d b]);
+     printf("\n");
+   }
+}' > out

report/benchmarks/bus-analytics/README.md

@@ -0,0 +1,9 @@
+# Mass-Transport System Analytics
+
+This set of scripts script is part of [a recent study on OASA](https://insidestory.gr/article/noymera-leoforeia-athinas) from Diomidis Spinellis and Eleftheria Tsaliki.  OASA is the the mass-transport system supporting the city of Athens. 
+
+1. `1.sh`: Vehicles on the road per day
+2. `2.sh`: Days a vehicle is on the road
+3. `3.sh`: Hours each vehicle is on the road
+4. `4.sh`: Hours monitored each day
+5. `5.sh`: Hours each bus is active each day

report/benchmarks/bus-analytics/full.sh

@@ -0,0 +1,61 @@
+#!/bin/bash
+
+# full bus analytics script
+
+# 1.sh Vehicles on the road per day
+cat $INPUT_FILE |
+  sed 's/T..:..:..//' |
+  cut -d ',' -f 1,3 |
+  sort -u |
+  cut -d ',' -f 1 |
+  sort |
+  uniq -c |
+  awk -v OFS="\t" "{print \$2,\$1}"
+
+# 2.sh Days a vehicle is on the road
+cat $INPUT_FILE |
+  sed 's/T..:..:..//' |
+  cut -d ',' -f 3,1 |
+  sort -u |
+  cut -d ',' -f 2 |
+  sort |
+  uniq -c |
+  sort -k1n |
+  awk -v OFS="\t" "{print \$2,\$1}"
+
+# 3.sh Hours each vehicle is on the road
+cat $INPUT_FILE |
+  sed 's/T\(..\):..:../,\1/' |
+  cut -d ',' -f 1,2,4 |
+  sort -u |
+  cut -d ',' -f 3 |
+  sort |
+  uniq -c |
+  sort -k1n |
+  awk -v OFS="\t" "{print \$2,\$1}"
+
+# 4.sh -- Hours monitored each day
+cat $INPUT_FILE |
+  sed 's/T\(..\):..:../,\1/' |
+  cut -d ',' -f 1,2 |
+  sort -u |
+  cut -d ',' -f 1 |
+  sort |
+  uniq -c |
+  awk -v OFS="\t" "{print \$2,\$1}"
+
+# 5.sh -- Records are day, hour, line, bus
+<$INPUT_FILE sed 's/T\(..\):..:../,\1/' | awk -F, '
+!seen[$1 $2 $4] { seen[$1 $2 $4] = 1; hours[$1 $4]++; bus[$4] = 1; day[$1] = 1; }
+END {
+   PROCINPUT_FILEFO["sorted_in"] = "@ind_str_asc"
+   for (d in day)
+     printf("\t%s", d);
+   printf("\n");
+   for (b in bus) {
+     printf("%s", b);
+     for (d in day)
+       printf("\t%s", hours[d b]);
+     printf("\n");
+   }
+}' > out