Table of Contents

• 1  Wikibooks SQL Exercise: The Warehouse
• 2  Load the libraries
• 3  Connect to database using postgres
• 4  Create Tables
• 5  Pandas and Spark dataframes
  • 5.1  data: Warehouses
  • 5.2  data: Boxes
  • 5.3  data: combined data
  • 5.4  merge two dataframes using pyspark
  • 5.5  spark dataframe attributes
• 6  Create SQL table from spark dataframes
• 7  Exercise Questions
  • 7.1  1. Select all warehouses.
  • 7.2  2. Select all boxes with a value larger than 150.
  • 7.3  3. Select all distinct contents in all the boxes.
  • 7.4  4. Select the average value of all the boxes.
  • 7.5  5. Select the warehouse code and the average value of the boxes in each warehouse.
  • 7.6  6. Same as previous exercise, but select only those warehouses where the average value of the boxes is greater than 150.
  • 7.7  7. Select the code of each box, along with the name of the city the box is located in.
  • 7.8  8. Select the warehouse codes, along with the number of boxes in each warehouse. Optionally, take into account that some warehouses are empty (i.e., the box count should show up as zero, instead of omitting the warehouse from the result).
  • 7.9  9. Select the codes of all warehouses that are saturated (a warehouse is saturated if the number of boxes in it is larger than the warehouse's capacity).
  • 7.10  10. Select the codes of all the boxes located in Chicago.
  • 7.11  11. Create a new warehouse in New York with a capacity for 3 boxes.
  • 7.12  12. Create a new box, with code "H5RT", containing "Papers" with a value of 200, and located in warehouse 2.
  • 7.13  13. Reduce the value of all boxes by 15%.
  • 7.14  14. Apply a 20% value reduction to boxes with a value larger than the average value of all the boxes.
  • 7.15  15. Remove all boxes with a value lower than 100.
  • 7.16  16. Remove all boxes from saturated warehouses.
• 8  Time Taken

Wikibooks SQL Exercise: The Warehouse

• https://en.wikibooks.org/wiki/SQL_Exercises
• https://en.wikibooks.org/wiki/SQL_Exercises/The_warehouse

Load the libraries

import time
time_start_notebook = time.time()

# my local library
import sys
sys.path.append("/Users/poudel/Dropbox/a00_Bhishan_Modules/")
sys.path.append("/Users/poudel/Dropbox/a00_Bhishan_Modules/bhishan")
from bhishan import bp

import numpy as np
import pandas as pd

import functools # reduce

import findspark, pyspark
%load_ext watermark
%watermark -a "Bhishan Poudel" -d -v -m
print()
%watermark -iv

Author: Bhishan Poudel

Python implementation: CPython
Python version       : 3.7.7
IPython version      : 7.16.1

Compiler    : Clang 4.0.1 (tags/RELEASE_401/final)
OS          : Darwin
Release     : 19.6.0
Machine     : x86_64
Processor   : i386
CPU cores   : 4
Architecture: 64bit


numpy    : 1.17.5
pyspark  : 3.1.2
bhishan  : 0.4
findspark: 1.4.2
sys      : 3.7.7 (default, May  6 2020, 04:59:01) 
[Clang 4.0.1 (tags/RELEASE_401/final)]
pandas   : 1.0.5

# pyspark settings
import os
import findspark

HOME = os.path.expanduser('~')
findspark.init(HOME + "/Softwares/Spark/spark-3.1.2-bin-hadoop3.2")

# Set spark environments
os.environ['PYSPARK_PYTHON'] = f'{HOME}/opt/miniconda3/envs/spk/bin/python'
os.environ['PYSPARK_DRIVER_PYTHON'] = f'{HOME}/opt/miniconda3/envs/spk/bin/python'

import pyspark
from pyspark import SparkConf, SparkContext, SQLContext
from pyspark.sql.session import SparkSession
from pyspark.sql.functions import col
from pyspark.sql.functions import col as _col
from pyspark.sql.functions import udf # @udf("integer") def myfunc(x,y): return x - y
from pyspark.sql import functions as F # stddev format_number date_format, dayofyear, when
from pyspark.sql.window import Window
from pyspark.sql.types import StructField, StringType, IntegerType,  FloatType, StructType
from pyspark.sql.functions import (mean as _mean, min as _min,
                                   max as _max, avg as _avg,
                                   when as _when
                                  )


spark = SparkSession\
        .builder\
        .appName("bhishan")\
        .getOrCreate()

sc = spark.sparkContext
sqlContext = pyspark.SQLContext(sc) # spark_df = sqlContext.createDataFrame(pandas_df)
sc.setLogLevel("INFO")
print(pyspark.__version__)

3.1.2

Connect to database using postgres

# postgres configs
import os, json
data = json.load(open(os.path.expanduser('~/.config/postgres/credentials.json')))
pw,port = data['password'], data['port'] # port is 5432
dbname = 'wbwarehouse' # make sure pgadmin is running and given database is there.

%load_ext sql
%sql postgres://postgres:$pw@localhost:$port/$dbname

Create Tables

%%sql

DROP TABLE IF EXISTS Warehouses CASCADE;


DROP TABLE IF EXISTS Boxes CASCADE;


CREATE TABLE Warehouses (Code INTEGER PRIMARY KEY NOT NULL,
                                                  LOCATION TEXT NOT NULL,
                                                                Capacity INTEGER NOT NULL);


CREATE TABLE Boxes (Code TEXT PRIMARY KEY NOT NULL,
                                          CONTENTS TEXT NOT NULL,
                                                        Value REAL NOT NULL,
                                                                   Warehouse INTEGER NOT NULL,
                                                                                     CONSTRAINT fk_Warehouses_Code
                    FOREIGN KEY (Warehouse) REFERENCES Warehouses(Code));


INSERT INTO Warehouses(Code, LOCATION, Capacity)
VALUES(1,'Chicago',3);


INSERT INTO Warehouses(Code, LOCATION, Capacity)
VALUES(2,'Chicago',4);


INSERT INTO Warehouses(Code, LOCATION, Capacity)
VALUES(3,'New York',7);


INSERT INTO Warehouses(Code, LOCATION, Capacity)
VALUES(4,'Los Angeles',2);


INSERT INTO Warehouses(Code, LOCATION, Capacity)
VALUES(5,'San Francisco',8);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('0MN7','Rocks',180,3);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('4H8P','Rocks',250,1);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('4RT3','Scissors',190,4);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('7G3H','Rocks',200,1);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('8JN6','Papers',75,1);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('8Y6U','Papers',50,3);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('9J6F','Papers',175,2);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('LL08','Rocks',140,4);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('P0H6','Scissors',125,1);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('P2T6','Scissors',150,2);


INSERT INTO Boxes(Code, CONTENTS, Value, Warehouse)
VALUES('TU55','Papers',90,5);


SELECT *
FROM Warehouses;

 * postgres://postgres:***@localhost:5432/wbwarehouse
Done.
Done.
Done.
Done.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
1 rows affected.
5 rows affected.

code	location	capacity
1	Chicago	3
2	Chicago	4
3	New York	7
4	Los Angeles	2
5	San Francisco	8

Pandas and Spark dataframes

data: Warehouses

x = %sql select * from Warehouses;

dfw = x.DataFrame()
dfw

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

	code	location	capacity
0	1	Chicago	3
1	2	Chicago	4
2	3	New York	7
3	4	Los Angeles	2
4	5	San Francisco	8

dfw.dtypes

code         int64
location    object
capacity     int64
dtype: object

schema = StructType([
    StructField('code',IntegerType(),True),
    StructField('location',StringType(),True),
    StructField('capacity',IntegerType(),True),
    ])

sdfw = sqlContext.createDataFrame(dfw, schema)
sdfw.show()

+----+-------------+--------+
|code|     location|capacity|
+----+-------------+--------+
|   1|      Chicago|       3|
|   2|      Chicago|       4|
|   3|     New York|       7|
|   4|  Los Angeles|       2|
|   5|San Francisco|       8|
+----+-------------+--------+

data: Boxes

x = %sql select * from Boxes;

dfb = x.DataFrame()
dfb

 * postgres://postgres:***@localhost:5432/wbwarehouse
11 rows affected.

	code	contents	value	warehouse
0	0MN7	Rocks	180.0	3
1	4H8P	Rocks	250.0	1
2	4RT3	Scissors	190.0	4
3	7G3H	Rocks	200.0	1
4	8JN6	Papers	75.0	1
5	8Y6U	Papers	50.0	3
6	9J6F	Papers	175.0	2
7	LL08	Rocks	140.0	4
8	P0H6	Scissors	125.0	1
9	P2T6	Scissors	150.0	2
10	TU55	Papers	90.0	5

dfb.dtypes

code          object
contents      object
value        float64
warehouse      int64
dtype: object

schema = StructType([
    StructField('code',StringType(),True),
    StructField('contents',StringType(),True),
    StructField('value',FloatType(),True),
    StructField('warehouse',IntegerType(),True),

    ])

sdfb = sqlContext.createDataFrame(dfb, schema)
sdfb.show()

+----+--------+-----+---------+
|code|contents|value|warehouse|
+----+--------+-----+---------+
|0MN7|   Rocks|180.0|        3|
|4H8P|   Rocks|250.0|        1|
|4RT3|Scissors|190.0|        4|
|7G3H|   Rocks|200.0|        1|
|8JN6|  Papers| 75.0|        1|
|8Y6U|  Papers| 50.0|        3|
|9J6F|  Papers|175.0|        2|
|LL08|   Rocks|140.0|        4|
|P0H6|Scissors|125.0|        1|
|P2T6|Scissors|150.0|        2|
|TU55|  Papers| 90.0|        5|
+----+--------+-----+---------+

data: combined data

df = dfw.merge(dfb,left_on='code', right_on='warehouse',suffixes=['_ware','_box'])
df

	code_ware	location	capacity	code_box	contents	value	warehouse
0	1	Chicago	3	4H8P	Rocks	250.0	1
1	1	Chicago	3	7G3H	Rocks	200.0	1
2	1	Chicago	3	8JN6	Papers	75.0	1
3	1	Chicago	3	P0H6	Scissors	125.0	1
4	2	Chicago	4	9J6F	Papers	175.0	2
5	2	Chicago	4	P2T6	Scissors	150.0	2
6	3	New York	7	0MN7	Rocks	180.0	3
7	3	New York	7	8Y6U	Papers	50.0	3
8	4	Los Angeles	2	4RT3	Scissors	190.0	4
9	4	Los Angeles	2	LL08	Rocks	140.0	4
10	5	San Francisco	8	TU55	Papers	90.0	5

merge two dataframes using pyspark

• we need to rename common columns of two tables.

cols_w = sdfw.columns
cols_b  = sdfb.columns
cols_common = set(cols_w) & set(cols_b)
cols_w, cols_b, cols_common

(['code', 'location', 'capacity'],
 ['code', 'contents', 'value', 'warehouse'],
 {'code'})

# rename common columns of first spark dataframe.

sdfw2 = sdfw.withColumnRenamed('code','code_ware')
sdfw2.show(2)

+---------+--------+--------+
|code_ware|location|capacity|
+---------+--------+--------+
|        1| Chicago|       3|
|        2| Chicago|       4|
+---------+--------+--------+
only showing top 2 rows

# rename common columns of second spark dataframe.

sdfb2 = sdfb.withColumnRenamed('code','code_box')
sdfb2.show(2)

+--------+--------+-----+---------+
|code_box|contents|value|warehouse|
+--------+--------+-----+---------+
|    0MN7|   Rocks|180.0|        3|
|    4H8P|   Rocks|250.0|        1|
+--------+--------+-----+---------+
only showing top 2 rows

# now join two spark dataframes which have different column names.

sdf = sdfw2.join(sdfb2,sdfw2.code_ware==sdfb2.warehouse)
sdf.show()

+---------+-------------+--------+--------+--------+-----+---------+
|code_ware|     location|capacity|code_box|contents|value|warehouse|
+---------+-------------+--------+--------+--------+-----+---------+
|        1|      Chicago|       3|    4H8P|   Rocks|250.0|        1|
|        1|      Chicago|       3|    7G3H|   Rocks|200.0|        1|
|        1|      Chicago|       3|    8JN6|  Papers| 75.0|        1|
|        1|      Chicago|       3|    P0H6|Scissors|125.0|        1|
|        3|     New York|       7|    0MN7|   Rocks|180.0|        3|
|        3|     New York|       7|    8Y6U|  Papers| 50.0|        3|
|        5|San Francisco|       8|    TU55|  Papers| 90.0|        5|
|        4|  Los Angeles|       2|    4RT3|Scissors|190.0|        4|
|        4|  Los Angeles|       2|    LL08|   Rocks|140.0|        4|
|        2|      Chicago|       4|    9J6F|  Papers|175.0|        2|
|        2|      Chicago|       4|    P2T6|Scissors|150.0|        2|
+---------+-------------+--------+--------+--------+-----+---------+

spark dataframe attributes

bp.show_methods(sdfw,4)
# count, fillna, filter, where, groubpy, first, corr,
# missing than pandas: max, min, sum, etc.

	0	1	2	3
0	agg	drop_duplicates	localCheckpoint	sql_ctx
1	alias	dropna	mapInPandas	stat
2	approxQuantile	dtypes	na	storageLevel
3	cache	exceptAll	orderBy	subtract
4	checkpoint	explain	persist	summary
5	coalesce	fillna	printSchema	tail
6	colRegex	filter	randomSplit	take
7	collect	first	rdd	toDF
8	columns	foreach	registerTempTable	toJSON
9	corr	foreachPartition	repartition	toLocalIterator
10	count	freqItems	repartitionByRange	toPandas
11	cov	groupBy	replace	transform
12	createGlobalTempView	groupby	rollup	union
13	createOrReplaceGlobalTempView	head	sameSemantics	unionAll
14	createOrReplaceTempView	hint	sample	unionByName
15	createTempView	inputFiles	sampleBy	unpersist
16	crossJoin	intersect	schema	where
17	crosstab	intersectAll	select	withColumn
18	cube	isLocal	selectExpr	withColumnRenamed
19	describe	isStreaming	semanticHash	withWatermark
20	distinct	is_cached	show	write
21	drop	join	sort	writeStream
22	dropDuplicates	limit	sortWithinPartitions	writeTo

Create SQL table from spark dataframes

• https://spark.apache.org/docs/2.2.0/sql-programming-guide.html

Global Temporary View

Temporary views in Spark SQL are session-scoped and will disappear if the session that creates it terminates. If you want to have a temporary view that is shared among all sessions and keep alive until the Spark application terminates, you can create a global temporary view. Global temporary view is tied to a system preserved database global_temp, and we must use the qualified name to refer it, e.g. SELECT * FROM global_temp.view1.

# Register the DataFrame as a SQL temporary view
sdfw.createOrReplaceTempView("Warehouses")
sdfb.createOrReplaceTempView("Boxes")
sdf.createOrReplaceTempView("sdf")

spark.sql('select * from Warehouses limit 2').show()

+----+--------+--------+
|code|location|capacity|
+----+--------+--------+
|   1| Chicago|       3|
|   2| Chicago|       4|
+----+--------+--------+

Exercise Questions

1. Select all warehouses.

%%sql

SELECT *
FROM Warehouses

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

code	location	capacity
1	Chicago	3
2	Chicago	4
3	New York	7
4	Los Angeles	2
5	San Francisco	8

2. Select all boxes with a value larger than 150.

# sdfb[sdfb['Value']>150].show()

%%sql

SELECT *
FROM Boxes
WHERE value > 150

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

code	contents	value	warehouse
0MN7	Rocks	180.0	3
4H8P	Rocks	250.0	1
4RT3	Scissors	190.0	4
7G3H	Rocks	200.0	1
9J6F	Papers	175.0	2

3. Select all distinct contents in all the boxes.

sdfb.select('Contents').distinct().show()

+--------+
|Contents|
+--------+
|   Rocks|
|  Papers|
|Scissors|
+--------+

%%sql

SELECT distinct(CONTENTS)
FROM Boxes

 * postgres://postgres:***@localhost:5432/wbwarehouse
3 rows affected.

contents
Scissors
Papers
Rocks

4. Select the average value of all the boxes.

sdfb.select(_mean('Value')).show()

+------------------+
|        avg(Value)|
+------------------+
|147.72727272727272|
+------------------+

%%sql

SELECT avg(value)
FROM Boxes

 * postgres://postgres:***@localhost:5432/wbwarehouse
1 rows affected.

avg
147.727272727273

5. Select the warehouse code and the average value of the boxes in each warehouse.

dfb.head(2)

	code	contents	value	warehouse
0	0MN7	Rocks	180.0	3
1	4H8P	Rocks	250.0	1

# dfb.groupby('warehouse')['value'].mean()

# dfb.groupby('warehouse').agg(mean_value=('value','mean'))

sdfb.groupby('warehouse').agg({'value':'mean'}).show()

+---------+----------+
|warehouse|avg(value)|
+---------+----------+
|        1|     162.5|
|        3|     115.0|
|        5|      90.0|
|        4|     165.0|
|        2|     162.5|
+---------+----------+

%%sql

SELECT warehouse,
       avg(value)
FROM Boxes
GROUP BY warehouse

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

warehouse	avg
3	115.0
5	90.0
4	165.0
2	162.5
1	162.5

6. Same as previous exercise, but select only those warehouses where the average value of the boxes is greater than 150.

dfb.groupby('warehouse')['value'].mean()[lambda x: x>150]

warehouse
1    162.5
2    162.5
4    165.0
Name: value, dtype: float64

# dfb.groupby('warehouse').filter(lambda dfx: dfx['value'].mean()>150)

sdfb.groupby('warehouse').agg({'value':'mean'}).filter('avg(value)>150').show()

+---------+----------+
|warehouse|avg(value)|
+---------+----------+
|        1|     162.5|
|        4|     165.0|
|        2|     162.5|
+---------+----------+

%%sql

SELECT warehouse,
       avg(value) AS avg_value
FROM Boxes
GROUP BY warehouse
HAVING avg(value)>150

 * postgres://postgres:***@localhost:5432/wbwarehouse
3 rows affected.

warehouse	avg_value
4	165.0
2	162.5
1	162.5

7. Select the code of each box, along with the name of the city the box is located in.

# dfb.head(2)

df.head(2)

	code_ware	location	capacity	code_box	contents	value	warehouse
0	1	Chicago	3	4H8P	Rocks	250.0	1
1	1	Chicago	3	7G3H	Rocks	200.0	1

# df[['code_box','location']]

# sdf.select('code_box','location').show()

%%sql

SELECT b.code AS code_box,
       w.location AS LOCATION
FROM Boxes b,
     Warehouses w
WHERE b.warehouse = w.code

 * postgres://postgres:***@localhost:5432/wbwarehouse
11 rows affected.

code_box	location
0MN7	New York
4H8P	Chicago
4RT3	Los Angeles
7G3H	Chicago
8JN6	Chicago
8Y6U	New York
9J6F	Chicago
LL08	Los Angeles
P0H6	Chicago
P2T6	Chicago
TU55	San Francisco

8. Select the warehouse codes, along with the number of boxes in each warehouse. Optionally, take into account that some warehouses are empty (i.e., the box count should show up as zero, instead of omitting the warehouse from the result).

# dfb

# dfb.groupby('warehouse')['code'].count()

# sdfb.groupby('warehouse').agg({'code':'count'}).show()

%%sql

SELECT warehouse,
       count(*)
FROM Boxes
GROUP BY warehouse

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

warehouse	count
3	2
5	1
4	2
2	2
1	4

%%sql
/* taking into account nulls */
SELECT w.code,
       count(w.code)
FROM Warehouses w
LEFT JOIN Boxes b ON w.code = b.warehouse
GROUP BY w.code

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

code	count
2	2
3	2
5	1
4	2
1	4

9. Select the codes of all warehouses that are saturated (a warehouse is saturated if the number of boxes in it is larger than the warehouse's capacity).

df['num_boxes_ware'] = df.groupby('warehouse')['code_box'].transform('count')
df.head(2)

	code_ware	location	capacity	code_box	contents	value	warehouse	num_boxes_ware
0	1	Chicago	3	4H8P	Rocks	250.0	1	4
1	1	Chicago	3	7G3H	Rocks	200.0	1	4

df[df.num_boxes_ware>df.capacity]['code_ware'].unique()

array([1])

# spark does not have groupby transform
# https://github.com/bhishanpdl/Tutorials_and_Lessons/blob/master/Tutorial_PySpark/z01_pyspark_methods/z01_pyspark_groupby/README.md

spark.sql("""
select code from Warehouses w
where capacity <
(
  select count(*) from Boxes b 
  where b.warehouse = w.code
)

""").show()

+----+
|code|
+----+
|   1|
+----+

# next cell command does not work in pyspark, only work in sql

%%sql

SELECT w.code
FROM Warehouses w
JOIN Boxes b ON w.code = b.warehouse
GROUP BY w.code
HAVING count(b.code) > w.capacity

 * postgres://postgres:***@localhost:5432/wbwarehouse
1 rows affected.

code
1

spark.sql("""
select w.code
from Warehouses w
join Boxes b
on w.code = b.warehouse
group by w.code
having count(b.code) > first(w.capacity)

""").show()

# NOTE: We need FIRST function inside HAVING clause.

+----+
|code|
+----+
|   1|
+----+

10. Select the codes of all the boxes located in Chicago.

# df[df.location=='Chicago']['code_box']

# sdf.filter(_col('location')=='Chicago').select('code_box').distinct().show()

%%sql

SELECT b.code
FROM Warehouses w
RIGHT JOIN Boxes b ON w.code=b.warehouse
WHERE w.location='Chicago'

 * postgres://postgres:***@localhost:5432/wbwarehouse
6 rows affected.

code
4H8P
7G3H
8JN6
9J6F
P0H6
P2T6

%%sql

SELECT *
FROM Boxes
LIMIT 1;

 * postgres://postgres:***@localhost:5432/wbwarehouse
1 rows affected.

code	contents	value	warehouse
0MN7	Rocks	180.0	3

%%sql

SELECT code
FROM Boxes
WHERE warehouse IN
    (SELECT code
     FROM Warehouses
     WHERE LOCATION='Chicago' )

 * postgres://postgres:***@localhost:5432/wbwarehouse
6 rows affected.

code
4H8P
7G3H
8JN6
9J6F
P0H6
P2T6

spark.sql("""
select b.code
from Warehouses w
right join Boxes b
on w.code=b.warehouse
where w.location='Chicago'
""").take(1)

[Row(code='4H8P')]

spark.sql("""
select code from Boxes
where warehouse in (
    select code from Warehouses where location='Chicago'
)
""").take(1)

[Row(code='4H8P')]

11. Create a new warehouse in New York with a capacity for 3 boxes.

%%sql

SELECT *
FROM Warehouses;

 * postgres://postgres:***@localhost:5432/wbwarehouse
5 rows affected.

code	location	capacity
1	Chicago	3
2	Chicago	4
3	New York	7
4	Los Angeles	2
5	San Francisco	8

%%sql

INSERT INTO Warehouses (code, LOCATION, Capacity)
VALUES (6, 'New York',3);

 * postgres://postgres:***@localhost:5432/wbwarehouse
1 rows affected.

[]

%%sql

SELECT *
FROM warehouses;

 * postgres://postgres:***@localhost:5432/wbwarehouse
6 rows affected.

code	location	capacity
1	Chicago	3
2	Chicago	4
3	New York	7
4	Los Angeles	2
5	San Francisco	8
6	New York	3

12. Create a new box, with code "H5RT", containing "Papers" with a value of 200, and located in warehouse 2.

%%sql

SELECT *
FROM warehouses;

 * postgres://postgres:***@localhost:5432/wbwarehouse
6 rows affected.

code	location	capacity
1	Chicago	3
2	Chicago	4
3	New York	7
4	Los Angeles	2
5	San Francisco	8
6	New York	3

%%sql

INSERT INTO Boxes
VALUES('H5RT','Papers',200,2);

 * postgres://postgres:***@localhost:5432/wbwarehouse
1 rows affected.

[]

%%sql

SELECT *
FROM warehouses;

 * postgres://postgres:***@localhost:5432/wbwarehouse
6 rows affected.

code	location	capacity
1	Chicago	3
2	Chicago	4
3	New York	7
4	Los Angeles	2
5	San Francisco	8
6	New York	3

13. Reduce the value of all boxes by 15%.

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
12 rows affected.

code	contents	value	warehouse
0MN7	Rocks	180.0	3
4H8P	Rocks	250.0	1
4RT3	Scissors	190.0	4
7G3H	Rocks	200.0	1
8JN6	Papers	75.0	1
8Y6U	Papers	50.0	3
9J6F	Papers	175.0	2
LL08	Rocks	140.0	4
P0H6	Scissors	125.0	1
P2T6	Scissors	150.0	2
TU55	Papers	90.0	5
H5RT	Papers	200.0	2

%%sql

UPDATE Boxes
SET Value = Value * 0.85;

 * postgres://postgres:***@localhost:5432/wbwarehouse
12 rows affected.

[]

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
12 rows affected.

code	contents	value	warehouse
0MN7	Rocks	153.0	3
4H8P	Rocks	212.5	1
4RT3	Scissors	161.5	4
7G3H	Rocks	170.0	1
8JN6	Papers	63.75	1
8Y6U	Papers	42.5	3
9J6F	Papers	148.75	2
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
TU55	Papers	76.5	5
H5RT	Papers	170.0	2

14. Apply a 20% value reduction to boxes with a value larger than the average value of all the boxes.

dfb

	code	contents	value	warehouse
0	0MN7	Rocks	180.0	3
1	4H8P	Rocks	250.0	1
2	4RT3	Scissors	190.0	4
3	7G3H	Rocks	200.0	1
4	8JN6	Papers	75.0	1
5	8Y6U	Papers	50.0	3
6	9J6F	Papers	175.0	2
7	LL08	Rocks	140.0	4
8	P0H6	Scissors	125.0	1
9	P2T6	Scissors	150.0	2
10	TU55	Papers	90.0	5

dfb['value'].mean()

147.72727272727272

dfb['discounted_value'] = np.where(dfb['value']>dfb['value'].mean(),
                                  dfb['value']*0.8,
                                  dfb['value'])

dfb

	code	contents	value	warehouse	discounted_value
0	0MN7	Rocks	180.0	3	144.0
1	4H8P	Rocks	250.0	1	200.0
2	4RT3	Scissors	190.0	4	152.0
3	7G3H	Rocks	200.0	1	160.0
4	8JN6	Papers	75.0	1	75.0
5	8Y6U	Papers	50.0	3	50.0
6	9J6F	Papers	175.0	2	140.0
7	LL08	Rocks	140.0	4	140.0
8	P0H6	Scissors	125.0	1	125.0
9	P2T6	Scissors	150.0	2	120.0
10	TU55	Papers	90.0	5	90.0

m = sdfb.select(F.mean('value')).collect()[0][0]
m

147.72727272727272

sdfb.withColumn('discounted_value',
                F.when( F.col('value') > m,
                        F.col('value')*0.8
                      )
                .otherwise(F.col('value'))
               ).show()

+----+--------+-----+---------+----------------+
|code|contents|value|warehouse|discounted_value|
+----+--------+-----+---------+----------------+
|0MN7|   Rocks|180.0|        3|           144.0|
|4H8P|   Rocks|250.0|        1|           200.0|
|4RT3|Scissors|190.0|        4|           152.0|
|7G3H|   Rocks|200.0|        1|           160.0|
|8JN6|  Papers| 75.0|        1|            75.0|
|8Y6U|  Papers| 50.0|        3|            50.0|
|9J6F|  Papers|175.0|        2|           140.0|
|LL08|   Rocks|140.0|        4|           140.0|
|P0H6|Scissors|125.0|        1|           125.0|
|P2T6|Scissors|150.0|        2|           120.0|
|TU55|  Papers| 90.0|        5|            90.0|
+----+--------+-----+---------+----------------+

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
12 rows affected.

code	contents	value	warehouse
0MN7	Rocks	153.0	3
4H8P	Rocks	212.5	1
4RT3	Scissors	161.5	4
7G3H	Rocks	170.0	1
8JN6	Papers	63.75	1
8Y6U	Papers	42.5	3
9J6F	Papers	148.75	2
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
TU55	Papers	76.5	5
H5RT	Papers	170.0	2

%%sql

UPDATE Boxes
SET Value = Value * 0.80
WHERE Value >
    (SELECT AVG(Value)
     FROM
       (SELECT *
        FROM Boxes) AS X);

 * postgres://postgres:***@localhost:5432/wbwarehouse
6 rows affected.

[]

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
12 rows affected.

code	contents	value	warehouse
8JN6	Papers	63.75	1
8Y6U	Papers	42.5	3
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
TU55	Papers	76.5	5
0MN7	Rocks	122.4	3
4H8P	Rocks	170.0	1
4RT3	Scissors	129.2	4
7G3H	Rocks	136.0	1
9J6F	Papers	119.0	2
H5RT	Papers	136.0	2

15. Remove all boxes with a value lower than 100.

# dfb[dfb['value']>100]

# sdfb[sdfb['value']>100].show()

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
12 rows affected.

code	contents	value	warehouse
8JN6	Papers	63.75	1
8Y6U	Papers	42.5	3
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
TU55	Papers	76.5	5
0MN7	Rocks	122.4	3
4H8P	Rocks	170.0	1
4RT3	Scissors	129.2	4
7G3H	Rocks	136.0	1
9J6F	Papers	119.0	2
H5RT	Papers	136.0	2

%%sql

DELETE
FROM Boxes
WHERE Value < 100;

 * postgres://postgres:***@localhost:5432/wbwarehouse
3 rows affected.

[]

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
9 rows affected.

code	contents	value	warehouse
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
0MN7	Rocks	122.4	3
4H8P	Rocks	170.0	1
4RT3	Scissors	129.2	4
7G3H	Rocks	136.0	1
9J6F	Papers	119.0	2
H5RT	Papers	136.0	2

16. Remove all boxes from saturated warehouses.

# df[df.num_boxes_ware > df.capacity]

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
9 rows affected.

code	contents	value	warehouse
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
0MN7	Rocks	122.4	3
4H8P	Rocks	170.0	1
4RT3	Scissors	129.2	4
7G3H	Rocks	136.0	1
9J6F	Papers	119.0	2
H5RT	Papers	136.0	2

%%sql

DELETE
FROM Boxes
WHERE Warehouse IN
    (SELECT Code
     FROM Warehouses
     WHERE Capacity <
         (SELECT COUNT(*)
          FROM Boxes
          WHERE Warehouse = Warehouses.Code ) );

 * postgres://postgres:***@localhost:5432/wbwarehouse
0 rows affected.

[]

%%sql

SELECT *
FROM Boxes;

 * postgres://postgres:***@localhost:5432/wbwarehouse
9 rows affected.

code	contents	value	warehouse
LL08	Rocks	119.0	4
P0H6	Scissors	106.25	1
P2T6	Scissors	127.5	2
0MN7	Rocks	122.4	3
4H8P	Rocks	170.0	1
4RT3	Scissors	129.2	4
7G3H	Rocks	136.0	1
9J6F	Papers	119.0	2
H5RT	Papers	136.0	2

Time Taken

time_taken = time.time() - time_start_notebook
h,m = divmod(time_taken,60*60)
print('Time taken to run whole notebook: {:.0f} hr '\
      '{:.0f} min {:.0f} secs'.format(h, *divmod(m,60)))

Time taken to run whole notebook: 0 hr 0 min 36 secs