Due at 11:59pm on 8/3/2017.

Starter Files

Download lab13.zip. Inside the archive, you will find starter files for the questions in this lab, along with a copy of the OK autograder.

Submission

By the end of this lab, you should have submitted the lab with python3 ok --submit. You may submit more than once before the deadline; only the final submission will be graded.

  • To receive credit for this lab, you must complete Questions 2-6 in lab13.sql, and submit through OK.
  • Question 7 and 8 are considered extra practice. They can be found in the lab13_extra.sql file. It is recommended that you complete them on your own time.

Setup

The simplest way to start using SQLite is to download a precompiled binary from the SQLite website. The latest version of SQLite at the time of writing is 3.15.1, but you can check for additional updates on the website.

Windows

  1. Visit the download page linked above and navigate to the section Precompiled Binaries for Windows. Click on the link sqlite-tools-win32-x86-*.zip to download the binary.
  2. Unzip the file. There should be a sqlite3.exe file in the directory after extraction.
  3. Navigate to the folder containing the sqlite3.exe file and check that the version is at least 3.8.3:

    $ cd path/to/sqlite
    $ ./sqlite3 --version
    3.12.1 2016-04-08 15:09:49 fe7d3b75fe1bde41511b323925af8ae1b910bc4d

macOS Yosemite (10.10), El Capitan (10.11), Sierra (10.12)

SQLite comes pre-installed. Check that you have a version that's greater than 3.8.3:

    $ sqlite3
    SQLite version 3.8.10.2

Mac OS X Mavericks (10.9) or older

SQLite comes pre-installed, but it is the wrong version.

  1. Visit the download page linked above and navigate to the section Precompiled Binaries for Mac OS X (x86). Click on the link sqlite-tools-osx-x86-*.zip to download the binary.
  2. Unzip the file. There should be a sqlite3 file in the directory after extraction.
  3. Navigate to the folder containing the sqlite3 file and check that the version is at least 3.8.3:

    $ cd path/to/sqlite
    $ ./sqlite3 --version
    3.12.1 2016-04-08 15:09:49 fe7d3b75fe1bde41511b323925af8ae1b910bc4d

Ubuntu

The easiest way to use SQLite on Ubuntu is to install it straight from the native repositories (the version will be slightly behind the most recent release):

$ sudo apt-get install sqlite3
$ sqlite3 --version
3.8.6 2014-08-15 11:46:33 9491ba7d738528f168657adb43a198238abde19e

Usage

Note: If you downloaded a precompiled binary above, make sure that sqlite3.exe file is in the same directory as your .sql file. (Extract and move it out from the zip file you downloaded.)

After writing your code in the .sql file, you can test and verify your output in Terminal or Git Bash with one of the two following commands.

1.) Runs your code and then exits SQLite immediately afterwards.

  • Ubuntu / Mac OS X (Yosemite or newer)

    sqlite3 < lab13.sql
  • Windows / Mac OS X (Mavericks or older)

    ./sqlite3 < lab13.sql

2.) Runs your code and keeps SQLite open for further commands, which is similar to running Python code with the interactive -i flag. You can type .help to see some of the commands you can run.

  • Ubuntu / Mac OS X (Yosemite or newer)

    sqlite3 --init lab13.sql
  • Windows / Mac OS X (Mavericks or older)

    ./sqlite3 --init lab13.sql

To exit out of SQLite after using the second command, type .exit or .quit or you can hit Ctrl-C. Also if you see ...> you probably forgot a ;

SQL Basics

Creating Tables

You can create SQL tables either from scratch or from existing tables.

Below creates the table from scratch, without referencing any other existing tables.

CREATE TABLE [table_name] AS
  SELECT [val1] AS [column1], [val2] AS [column2], ... UNION
  SELECT [val3]             , [val4]             , ... UNION
  SELECT [val5]             , [val6]             , ...;

Note: You do not need to repeat the AS keyword in subsequent SELECT statements when creating the table.

Here is an example where we construct a table with the CREATE TABLE statement. UNION is used here to join rows and AS assigns a table column to a new name.

CREATE TABLE Football AS
  SELECT 30 AS Berkeley, 7 AS Stanford, 2002 AS Year UNION
  SELECT 28,             16,            2003         UNION
  SELECT 17,             38,            2014;

football

Here we have created a table called Football, which has three attributes (columns): Berkeley, Stanford, and Year. We can later access the values from this table by referencing the table's columns.

To create tables from existing tables, the SELECT command references another table.

Selecting From Tables

More commonly, we will create new tables by selecting specific columns that we want from existing tables. SELECT statements can include optional clauses such as:

  • FROM: tells SQL which tables to select values from
  • WHERE: filters by some condition
  • ORDER BY: enforces an ordering by some attribute or attributes (usually a column or columns)
  • LIMIT: limits the number of rows in the output table

    SELECT [columns] FROM [tables] WHERE [condition] ORDER BY [attributes] LIMIT [limit];

Notes about the arguments:

  • [columns]: a comma separated list of the columns to select, * can be used to select all of them
  • [tables]: a comma separated list of tables to select values from
  • [condition]: a Boolean expression
  • [attributes]: a comma separated list of attributes, which are usually columns, but could also be named aggregates (which we will learn later)
  • [limit]: an integer

We can select all the values of an attribute from a table with the SELECT statement. In addition, we can apply a filter using the WHERE clause. Here, we filter by Year > 2002, which makes the SELECT statement keep only the rows in the table whose Year value is greater than 2002.

sqlite> SELECT Berkeley FROM Football WHERE Year > 2002;
17
28

Here we selected Berkeley's score for all years after 2002.

Expressions in SQL

Here are some fundamental operations you can perform:

  • comparisons: =, >, <, <=, >=, <> ("not equal")
  • booleans: AND, OR
  • arithmetic: +, -, *, /

We can also perform string concatenation using ||, which behaves similarly to + on strings in Python.

sqlite> SELECT "hello" || " " || "world";
hello world

Note we capitalize SQL syntax purely for style. It makes it much easier to read, though will work if you don't capitalize it.

Getting to Know Your Fellow 61A Students

In the past week, we asked you and your fellow students to complete a brief online survey through Google Forms, which involved relatively random but fun questions. In this lab, we will interact with the results of the survey by using SQL queries to see if we can find interesting things in the data.

First, take a look at su17data.sql and examine the table defined in it. Note its structure. You will be working with:

  • students: The main results of the survey. Each column represents a different question from the survey, except for the first column, which is the time of when the result was submitted. This time is a unique identifier for each of the rows in the table.

    Column Name Question
    time The unique timestamp that identifies the submission
    number What's your favorite number between 1 and 100?
    color What is your favorite color?
    seven Choose the number 7 below.
    Options:
    • 7
    • You are not the boss of me!
    • I do what I want.
    • I'm a rebel
    • Choose this option instead.
    • YOLO!
    • the number 7
    sneeze If all of a sudden all humans simultaneously lost the ability to sneeze, how long do you think it would take mankind as a collective to realize? (in hours)
    expert What do you want to be an expert in?
    beets If every rapper became a farmer, who would have the freshest BEETS?
    Options:
    • The Roots
    • 60 Cent
    • Post Melon
    • Tyler, the Cultivator
    • Metro' Bloomin (even though he's not a rapper)
    • 21 Cabbage
    • DJ Kale-ed
    • Kanye Fresh
    • Beets by Dre
    • Pea Diddy
    • Kendrick Farmar
    • 2 Grainz
    image Choose your favorite image. (Options shown under Question 7)
    pet If you could have any animal in the world as a pet, what would it be?
    date Pick a day of the year!
    smallest Try to guess the smallest unique positive INTEGER that anyone will put!
  • checkboxes: The results from the survey in which students could select more than one option from the numbers listed, which ranged from 0 to 10 and included 2017, 9000, and 9001. Each row has a time (which is again a unique identifier) and has the value 'True' if the student selected the column or 'False' if the student did not. The column names in this table are the following strings, referring to each possible number: '0', '1', '2', '4', '5', '6', '7', '8', '9', '10', '2016', '9000', '9001'.

A time in students matches up with a time in checkboxes. For example, the row with time "4/12/2017 19:12:04" in students matches up with the row with time "4/12/2017 19:12:04" in checkboxes. These entries come from the same Google form submission and thus belong to the same student. We used time to uniquely identify each student, rather than using their name or email.


Note: If you are looking for your personal response within the data, you may have noticed that some of your answers are slightly different from what you inputted. In order to make SQLite accept our data, and to optimize for as many matches as possible during our joins, we did the following things to clean up the data:

  • color and pet: We converted all the strings to be completely lowercase.
  • Blank answers were replaced with "n/a" or 404, depending on if the data was a string or number

You will write all of your solutions in the starter file lab13.sql provided. As with other labs, you can test your solutions with OK. In addition, you can use either of the following commands. You may need to refer to the Usage section to find the appropriate command for your OS:

sqlite3 < lab13.sql
sqlite3 --init lab13.sql

Question 1: What Would SQL print?

First, load the tables into sqlite3. If you're on Windows or Mac OS X (Mavericks or older), use the following command:

$ ./sqlite3 --init lab13.sql

If you're on Ubuntu or Mac OS X (Yosemite or newer), use:

$ sqlite3 --init lab13.sql

Before we start, inspect the schema of the tables that we've created for you:

sqlite> .schema

This tells you the name of each of our tables and their attributes.

Let's also take a look at some of the entries in our table. There are a lot of entries though, so let's just output the first 20:

sqlite> SELECT * FROM students LIMIT 20;

If you're curious about some of the answers students put into the Google form, open up su17data.sql in your favorite text editor and take a look!

For each of the SQL queries below, think about what the query is looking for, then try running the query yourself and see!

sqlite> SELECT * FROM students; -- This is a comment. * is shorthand for all columns!
______
selects all records from students;
sqlite> SELECT color FROM students WHERE number = 16;
______
selects the color from students who said their favorite number was 16;
sqlite> SELECT beets, pet FROM students WHERE color = "blue" AND date = "12/25";
______
selects the beet farmer and pet from students who said their favorite color was blue and picked December 25th;

Remember, to exit out of SQLite after using the second command, type .exit or .quit or hit Ctrl-C. Also, if you see ...> you probably forgot a ;

Question 2: Obedience

To warm-up, let's ask a simple question related to our data: Is there a correlation between whether students do as they're told and their favorite image?

stuff

Write a SQL query to create a table that contains the columns seven (this column representing "obedience") and image (the image students selected) from the students table.

You should get the following output:

sqlite> SELECT * FROM obedience LIMIT 10;
I do what I want.|Image 2
Choose this option instead.|Image 4
YOLO!|Image 1
I do what I want.|Image 4
the number 7|Image 3
Choose this option instead.|Image 1
7|Image 4
the number 7|Image 3
Choose this option instead.|Image 2
the number 7|Image 4
CREATE TABLE obedience AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT seven, image FROM students;

Use OK to test your code:

python3 ok -q obedience

Question 3: The Smallest Unique Integer

Who successfully managed to guess the smallest unique integer value? Let's find out!

Unfortunately we have not learned how to do aggregations, which can help us count the number of times a specific value was selected, in SQL just yet. As such, we can only hand inspect our data to determine it. However, an anonymous elf has informed us that the smallest unique value is greater than 5!

Write a SQL query with the columns time and smallest to try to determine what the smallest integer value is. In order to make it easier for us to inspect these values, use WHERE to restrict the answers to numbers greater than 5, ORDER BY to sort the numerical values, and LIMIT your result to the first 20 values that are greater than the number 5.

The first 5 lines of your output should look like this:

sqlite> SELECT * FROM smallest_int LIMIT 5;
7/30/2017 12:19:49|6
7/30/2017 12:20:11|6
7/30/2017 12:30:19|7
7/30/2017 12:53:09|7
7/30/2017 19:55:58|7
CREATE TABLE smallest_int AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT time, smallest FROM students WHERE smallest > 5 ORDER BY smallest LIMIT 20;

Use OK to test your code:

python3 ok -q smallest-int

After you've successfully passed the OK test, actually take a look at the table smallest_int that you just created and find the smallest unique integer value! If you're curious how to do this with aggregations, check the Extra Question.

To do this, try the following:

$ sqlite3 --init lab13.sql
sqlite> SELECT * FROM smallest_int; -- No LIMIT this time!

Joins

We can use joins to include rows from another table that satisfy the WHERE predicate. Joins can either be on different tables, or the same table if we include an alias. Here we are referencing the football table twice, using AS to bind the football table once as the alias a and once as the alias b.

sqlite> SELECT b.Berkeley - a.Berkeley, b.Stanford - a.Stanford, a.Year, b.Year
...>        FROM Football AS a, Football AS b WHERE a.Year < b.Year;
-11|22|2003|2014
-13|21|2002|2014
-2|9|2002|2003

What is this query asking for?

You may notice that it does not seem like we actually performed any operations to do the join. However, the join is implicit in the fact that we listed more than one table after the FROM. In this example, we joined the table Football with itself and gave each instance of the table an alias, a and b so that we could distinctly refer to each table's attributes and perform selections and comparisons on them, such as a.Year < b.Year.

One way to think of a join is that it produces a cross-product between the two tables by matching each row from the first table with every other row in the second table, which creates a new, larger joined table.

Here's an illustration of what happened in the joining process during the above query.

joins

From here, the select statement examines the joined table and selects the values it desires: b.Berkeley - a.Berkeley and b.Stanford - a.Stanford but only from the rows WHERE a.Year < b.Year. This prevents duplicate results from appearing in our output and also removes rows where the years are the same.

Question 4: Great Students Think Alike

We administered a similar survey for Lab 13 last semester to 61A students (see the table sp17students in sp17data.sql). The only questions that have changed are sneeze, expert, beets and image, so let's play around with this data using joins!

Great students think alike. We want to print out the information of students that have the same favorite date, color and pet! How do their favorite numbers differ?

Hint: When joining table names where column names are the same, use dot notation to distinguish which columns are from which table. The structure is [table_name].[column name]. This sometimes may get verbose, so it’s stylistically better to give tables an alias using the AS keyword. The structure is: SELECT <[alias1].[column name1], [alias2].[columnname2]...> from <[table_name1] as [alias1],[table_name2] as [alias2]...> ...

The query in the football example from earlier uses this syntax.

Write a SQL query to create a table that has 5 columns:

  • The shared favorite date between semesters
  • The shared favorite color between semesters
  • The shared favorite pet between semesters
  • The favorite number of this semester's student
  • The favorite number of last semester's student

You should get the following output:

sqlite> SELECT * FROM greatstudents;
3/4|blue|dragon|34|4
CREATE TABLE greatstudents AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT a.date, a.color, a.pet, a.number, b.number FROM students AS a, sp17students AS b WHERE a.date = b.date AND a.color = b.color AND a.pet = b.pet;

Use OK to test your code:

python3 ok -q greatstudents

After you've successfully passed the OK test, play around some more with joins to see what cool connections you can draw between the data from both semesters!

Question 5: Sevens

Let's take a look at data from both of our tables, students and checkboxes, to find out if students that really like the number 7 also chose '7' for the obedience question. Specifically, we want to look at the students that fulfill the below conditions and see if they also chose '7' in the question that asked students to choose the number 7 (column seven in students).

  • reported that their favorite number (column number in students) was 7
  • have 'True' in column '7' in checkboxes, meaning they checked the number 7 during the survey

In order to examine rows from both the students and the checkboxes table, we will need to perform a join.

How would you specify the WHERE clause to make the SELECT statement only consider rows in the joined table whose values all correspond to the same student? If you find that your output is massive and overwhelming, then you are probably missing the necessary condition in your WHERE clause to ensure this.

Note: The columns in the checkboxes table are strings with the associated number, so you must put quotes around the column name to refer to it. For example if you alias the table as a, to get the column to see if a student checked 9001, you must write a.'9001'.

Write a SQL query to create a table with just the column seven from students, filtering first for students who said their favorite number (column number) was 7 in the students table and who checked the box for seven (column '7') in the checkboxes table.

You should get the following output:

sqlite> SELECT * FROM sevens;
the number 7
the number 7
CREATE TABLE sevens AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT s.seven FROM students AS s, checkboxes AS c WHERE s.number = 7 AND c.'7' = 'True' AND s.time = c.time;

Use OK to test your code:

python3 ok -q sevens

Question 6: Matchmaker, Matchmaker

Did you take 61A with the hope of finding your soul mate? Well you're in luck! With all this data in hand, it's easy for us to find your perfect match. If two students want the same pet and have the same taste in beet producer, they are clearly meant to be together! In order to provide some more information for the potential lovebirds to converse about, let's include the favorite colors of the two individuals as well!

In order to match up students, you will have to do a join on the students table with itself. When you do a join, SQLite will match every single row with every single other row, so make sure you do not match anyone with themselves, or match any given pair twice!

Important Note: When pairing the first and second person, make sure that the first person responded first (i.e. they have an earlier time). This is to ensure your output matches our tests.

Hint: Remember you only want a pair of students to appear once. Refer back to the football example on how that was done.

Write a SQL query to create a table that has 4 columns:

  • The shared preferred pet of the couple
  • The shared favorite beets of the couple
  • The favorite color of the first person
  • The favorite color of the second person

You should get the following output:

sqlite> SELECT * FROM matchmaker;
dog|Kendrick Farmar|blue|white
lion|Beets by Dre|blue|yellow
dog|Tyler, the Cultivator|yellow|blue
dolphin|The Roots|turquoise|purple
dog|The Roots|yellow|grey
CREATE TABLE matchmaker AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT a.pet, a.beets, a.color, b.color FROM students AS a, students AS b WHERE a.time < b.time AND a.pet = b.pet AND a.beets = b.beets;

Use OK to test your code:

python3 ok -q matchmaker

Extra Questions

The following questions are for extra practice -- they can be found in the lab13_extra.sql file. It is recommended that you complete these problems, but you do not need to turn them in for credit.

The COUNT Aggregator

How many people liked each pet? What is the biggest date chosen this semester? How many obedient people chose Image 1? Is there a difference between last semester's average favorite number and this semester's?

To answer these types of questions, we can bring in SQL aggregation, which allows us to accumulate values across rows in our SQL database!

In order to perform SQL aggregation, we can group rows in our table by one or more attributes. Once we have groups, we can aggregate over the groups in our table and find things like:

  • the maximum value (MAX),
  • the minimum value (MIN),
  • the number of rows in the group (COUNT),
  • the average over all of the values (AVG),

and more! SELECT statements that use aggregation are usually marked by two things: an aggregate function (MAX, MIN, COUNT, AVG, etc.) and a GROUP BY clause. GROUP BY [column(s)] groups together rows with the same value in each column(s). In this section we will only use COUNT, which will count the number of rows in each group, but feel free to check out http://www.sqlcourse2.com/agg_functions.html for more!

For example, the following query will print out the top 10 favorite numbers with their respective counts:

sqlite> SELECT number, COUNT(*) AS count FROM students GROUP BY number
  ORDER BY count DESC LIMIT 10;
  42|8
  7|5
  4|4
  8|4
  27|4
  1|3
  5|3
  9|3
  11|3
  13|3

This SELECT statement first groups all of the rows in our table students by number. Then, within each group, we perform aggregation by COUNTing all the rows. By selecting number and COUNT(*), we then can see the highest number and how many students picked that number. We have to order by our COUNT(*), which is saved in the alias count, by DESCending order, so our highest count starts at the top, and we limit our result to the top 10.

Question 7: Let's Count

Let's have some fun with this! For each statement we created a separate table in lab13_extra.sql so fill in the corresponding table and run it using OK. Try working on this on your own or with a neighbor before toggling to see the solutions.

What was the favorite number from last semester?

sqlite> SELECT * from sp17favnum;
7|25
CREATE TABLE sp17favnum AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT number, COUNT(*) AS count FROM sp17students GROUP BY number ORDER BY count DESC LIMIT 1;

What were the top 10 pets last semester?

sqlite> SELECT * from sp17favpets;
dragon|20
dog|14
cat|8
unicorn|7
n/a|6
panda|5
dolphin|4
tiger|4
7|3
bear|3
CREATE TABLE sp17favpets AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT pet, COUNT(*) AS count FROM sp17students GROUP BY pet ORDER BY count DESC LIMIT 10;

What are the top 10 pets this semester?

sqlite> SELECT * from su17favpets;
dog|10
lion|5
cat|3
dolphin|3
dragon|3
tiger|3
panda|2
penguin|2
a bird|1
a dog|1
CREATE TABLE su17favpets AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT pet, COUNT(*) AS count FROM students GROUP BY pet ORDER BY count DESC LIMIT 10;

How many people marked exactly the word 'dog' as their ideal pet this semester?

sqlite> SELECT * from su17dog;
dog|10
CREATE TABLE su17dog AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT pet, COUNT(*) FROM students WHERE pet = 'dog';

Although close, our query doesn't give us an entirely accurate picture of what people's favorite pets are. For example, a dog would not be counted the same as dog. Let's see how many people actually want a dog this semester by using LIKE, which compares substrings. We can use it inside WHERE, as in WHERE [column_name] LIKE '%[word]%' to find how many people would like some type of dog.

sqlite> SELECT * from su17alldogs;
doggo|12
CREATE TABLE su17alldogs AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT pet, COUNT(*) FROM students WHERE pet LIKE '%dog%';

We can find the student's favorite for any column (try it yourself in the interpreter), but let's go back to our Obedience question. Let's see how many obedient students this semester picked each image. We can do this by selecting only the rows that have seven = '7' then GROUP BY image, and finally we can COUNT them.

sqlite> SELECT * from obedienceimage;
7|Image 1|15
7|Image 2|3
7|Image 3|5
7|Image 4|7
CREATE TABLE obedienceimage AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT seven, image, COUNT(*) FROM students WHERE seven = '7' GROUP BY image;

The possibilities are endless, so have fun experimenting!

Use OK to test your code:

python3 ok -q lets-count

Question 8: The Smallest Unique Integer (Part 2)

Now, let's revisit the previous problem of finding the smallest integer that anyone chose, and take a closer look at the COUNT aggregate.

Write a SQL query that uses the COUNT aggregate to create a table that pairs the attribute smallest with the number of times it was chosen by a student (this is the aggregation part).

Hint: You may find that there isn't a particular attribute you should have to perform the COUNT aggregation over. If you are only interested in counting the number of rows in a group, you can just say COUNT(*).

Hint: Think about what attribute you need to GROUP BY.

After you've defined your table, you should get something like:

sqlite> SELECT * FROM smallest_int_count LIMIT 25;
0|2
1|22
2|6
3|2
4|5
5|3
6|2
7|5
8|4
9|3
12|1
13|2
14|5
15|1
17|2
18|1
19|3
22|1
26|1
29|2
31|4
37|1
39|2
40|1
43|1
CREATE TABLE smallest_int_count AS
select "REPLACE THIS LINE WITH YOUR SOLUTION";
SELECT smallest, COUNT(*) FROM students GROUP BY smallest;

Use OK to test your code:

python3 ok -q smallest-int-count

It looks like the number 12 only had one person choose it! Were you the lucky student that put it down?