Assignment 01 - Jason Fan

__scripts__/bootstrap_data.sh

@@ -4,7 +4,7 @@ set -e
 set -x
 
 SCRIPTDIR=$(readlink -f $(dirname $0))
-DATADIR=$(readlink -f $(dirname $0)/../__data__)
+DATADIR=$(readlink -f $(dirname $0)/__data__)
 mkdir -p ${DATADIR}
 
 # Download and unzip data (if not already downloaded)

__scripts__/create_trip_tables.sql

@@ -1,3 +1,4 @@
+-- Active: 1770507353299@@localhost@5432@assignment1
 create schema if not exists indego;
 
 drop table if exists indego.trips_2021_q3;
@@ -20,6 +21,12 @@ create table indego.trips_2021_q3 (
     bike_type text
 );
 
+set DateStyle = 'MDY';
+copy indego.trips_2021_q3
+from 'F:/Graduate School Stuff/Semester 4/Deep Learning/assignment01/__data__/indego-trips-2021-q3.csv'
+with (format csv, header true);
+
+
 drop table if exists indego.trips_2022_q3;
 
 create table indego.trips_2022_q3 (
@@ -40,4 +47,9 @@ create table indego.trips_2022_q3 (
     bike_type text
 );
 
+set DateStyle = 'MDY';
+copy indego.trips_2022_q3
+from 'F:/Graduate School Stuff/Semester 4/Deep Learning/assignment01/__data__/indego-trips-2022-q3.csv'
+with (format csv, header true);
+
 create extension if not exists postgis;

__scripts__/script

@@ -0,0 +1,9 @@
+ogr2ogr \
+-f "PostgreSQL" \
+-nln "stations_geo" \
+-lco "SCHEMA=indego" \
+-lco "GEOM_TYPE=geography" \
+-lco "GEOMETRY_NAME=geog" \
+-lco "OVERWRITE=yes" \
+PG:"host=localhost port=5432 dbname=assignment1 user=postgres password='P123i,./g456'" \
+"/f/Graduate School Stuff/Semester 4/Deep Learning/assignment01/__data__/station_status.json"

compose.yml

@@ -2,14 +2,14 @@
 services:
   db:
     image: postgis/postgis
+    ports:
+      - "5432:5432"
     environment:
-      POSTGRES_PASSWORD: postgres
+      POSTGRES_PASSWORD: P123i,./g456
+      POSTGRES_DB: assignment1
     healthcheck:
       # test: ["CMD-SHELL", "pg_isready -U postgres"]
-      test: [
-          "CMD-SHELL",
-          'psql -U postgres -d postgres -c "SELECT ST_Buffer( ST_SetSRID(''POINT(0 0)''::GEOMETRY, 4326), 1) AS geom ;"',
-        ]
+      test: [ "CMD-SHELL", 'psql -U postgres -d postgres -c "SELECT ST_Buffer( ST_SetSRID(''POINT(0 0)''::GEOMETRY, 4326), 1) AS geom ;"' ]
       interval: 5s
       timeout: 5s
       retries: 5
@@ -18,10 +18,7 @@ services:
     image: ghcr.io/weitzman-musa-geocloud/assignment01:main
     volumes:
       - .:/assignment01
-    command: [
-      "bash", "-c",
-      "cd /assignment01 && __entrypoints__/run_linter.sh -- ."
-    ]
+    command: [ "bash", "-c", "cd /assignment01 && __entrypoints__/run_linter.sh -- ." ]
 
   run-tests:
     image: ghcr.io/weitzman-musa-geocloud/assignment01:main
@@ -36,7 +33,4 @@ services:
       POSTGRES_NAME: postgres
     volumes:
       - .:/assignment01
-    command: [
-      "bash", "-c",
-      "cd /assignment01 && npm install > /dev/null && __entrypoints__/run_tests.sh -- --rootDir /assignment01"
-    ]
+    command: [ "bash", "-c", "cd /assignment01 && npm install > /dev/null && __entrypoints__/run_tests.sh -- --rootDir /assignment01" ]

query02.sql

@@ -10,7 +10,16 @@
 */
 
 -- Enter your SQL query here
-
+select
+    round(
+        (count_2022 - count_2021) * 100.0 / count_2021,
+        2
+    )::text || '%' as perc_change
+from (
+    select
+        (select count(*) from indego.trips_2021_q3) as count_2021,
+        (select count(*) from indego.trips_2022_q3) as count_2022
+) as counts
 
 
 /*

query03.sql

@@ -6,3 +6,5 @@
 */
 
 -- Enter your SQL query here
+select round(avg(duration), 2) as avg_duration
+from indego.trips_2021_q3

query04.sql

@@ -6,3 +6,5 @@
 */
 
 -- Enter your SQL query here
+select round(avg(duration), 2) as avg_duration
+from indego.trips_2022_q3

query05.sql

@@ -5,3 +5,9 @@
 */
 
 -- Enter your SQL query here
+select max(duration) as max_duration
+from (
+    select duration from indego.trips_2021_q3
+    union all
+    select duration from indego.trips_2022_q3
+) as all_trips

query06.sql

@@ -7,3 +7,20 @@
 */
 
 -- Enter your SQL query here
+select
+    extract(year from start_time)::integer as trip_year,
+    extract(quarter from start_time)::integer as trip_quarter,
+    count(*) as num_trips
+from indego.trips_2021_q3
+where duration < 10
+group by trip_year, trip_quarter
+
+union all
+
+select
+    extract(year from start_time)::integer as trip_year,
+    extract(quarter from start_time)::integer as trip_quarter,
+    count(*) as num_trips
+from indego.trips_2022_q3
+where duration < 10
+group by trip_year, trip_quarter

query07.sql

@@ -6,7 +6,23 @@
 */
 
 -- Enter your SQL query here
+select
+    extract(year from start_time)::integer as trip_year,
+    extract(quarter from start_time)::integer as trip_quarter,
+    count(*) as num_trips
+from indego.trips_2021_q3
+where start_time::date != end_time::date
+group by trip_year, trip_quarter
 
+union all
+
+select
+    extract(year from start_time)::integer as trip_year,
+    extract(quarter from start_time)::integer as trip_quarter,
+    count(*) as num_trips
+from indego.trips_2022_q3
+where start_time::date != end_time::date
+group by trip_year, trip_quarter
 
 
 /*

query08.sql

@@ -9,6 +9,19 @@
 */
 
 -- Enter your SQL query here
+select
+    start_station as station_id,
+    st_point(start_lon, start_lat)::geography as stations_geog,
+    count(*) as num_trips
+from (
+    select * from indego.trips_2021_q3
+    union all
+    select * from indego.trips_2022_q3
+) as all_trips
+where extract(hour from start_time) between 7 and 9
+group by start_station, start_lon, start_lat
+order by num_trips desc
+limit 5
 
 
 /*

query09.sql

@@ -7,3 +7,13 @@
 */
 
 -- Enter your SQL query here
+select
+    passholder_type,
+    count(*) as num_trips
+from (
+    select passholder_type from indego.trips_2021_q3
+    union all
+    select passholder_type from indego.trips_2022_q3
+) as all_trips
+group by passholder_type
+order by num_trips desc

query10.sql

@@ -8,3 +8,9 @@
 */
 
 -- Enter your SQL query here
+select
+    id as station_id,
+    geog as station_geog,
+    round(st_distance(geog, st_point(-75.192584, 39.952415)::geography) / 50) * 50 as distance
+from indego.stations_geo
+order by station_id

query11.sql

@@ -5,3 +5,5 @@
 */
 
 -- Enter your SQL query here
+select round(avg(st_distance(geog, st_point(-75.192584, 39.952415)::geography)) / 1000) as avg_distance_km
+from indego.stations_geo

query12.sql

@@ -6,3 +6,6 @@
 */
 
 -- Enter your SQL query here
+select count(*) as num_stations
+from indego.stations_geo
+where st_distance(geog, st_point(-75.192584, 39.952415)::geography) <= 1000

query13.sql

@@ -7,3 +7,10 @@
 */
 
 -- Enter your SQL query here
+select
+    id as station_id,
+    name as station_name,
+    round(st_distance(geog, st_point(-75.192584, 39.952415)::geography) / 50) * 50 as distance
+from indego.stations_geo
+order by distance desc
+limit 1

query14.sql

@@ -7,3 +7,10 @@
 */
 
 -- Enter your SQL query here
+select
+    id as station_id,
+    name as station_name,
+    round(st_distance(geog, st_point(-75.192584, 39.952415)::geography) / 50) * 50 as distance
+from indego.stations_geo
+order by distance asc
+limit 1

script_postgis.sql

@@ -0,0 +1 @@
+CREATE EXTENSION postgis;