International Computer Institute, Izmir, Turkey SQL Asst.Prof.Dr.İlker Kocabaş UBİ502 at

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.2 of 48 UBI 502 Database Management Systems Overview Joins: types and conditions Example 1: our languages  relational algebra  tuple relational calculus  domain relational calculus  SQL Examples 2, 3  joins, aggregates, MySQL Join tricks

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.3 of 48 UBI 502 Database Management Systems Schema Diagram for the Banking Enterprise

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.4 of 48 UBI 502 Database Management Systems Joins Join types:  Inner join  Left outer join  Right outer join  Full outer join Join conditions:  Natural  On  Using (A 1, …, A n )

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.5 of 48 UBI 502 Database Management Systems The loan and borrower Relations

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.6 of 48 UBI 502 Database Management Systems loan INNER JOIN borrower ON loan.loan-number = borrower.loan-number

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.7 of 48 UBI 502 Database Management Systems loan LEFT OUTER JOIN borrower USINGloan-number) loan LEFT OUTER JOIN borrower USING (loan-number)

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.8 of 48 UBI 502 Database Management Systems Join Conditions a INNER JOIN b USING (c 1,..., c n ) SELECT * FROM loan INNER JOIN borrower USING (loan_number); | loan_number | branch_name | amount | customer_name | loan_number | | L-16 | Perryridge | 1300 | Adams | L-16 | | L-93 | Mianus | 500 | Curry | L-93 |... | L-17 | Downtown | 1000 | Williams | L-17 | Equivalent: a INNER JOIN b ON (a.c 1 = b.c 1, …, a.c n = b.c n )

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.9 of 48 UBI 502 Database Management Systems Join Conditions Suppose (c 1, …, c n ) is a complete list of attributes common to A and B Then: a INNER JOIN b USING (c 1, …, c n ) is equivalent to: a NATURAL JOIN b E.g. since there is only one common attribute between loan and borrower, the following queries give the same result: SELECT * FROM loan INNER JOIN borrower USING (loan_number); SELECT * FROM loan NATURAL JOIN borrower;

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.10 of 48 UBI 502 Database Management Systems Example 1: Our languages Query: What branches are in Brooklyn?  branch-name (  branch-city = “Brooklyn” (branch)) {t |  s  branch (t[branch-name] = s[branch-name]  s [branch-city] = “Brooklyn”)} {  n  |  c, a (  n, c, a   branch  c = “Brooklyn”)} SELECT branch-name FROM branch WHERE branch_city = “Brooklyn”

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.11 of 48 UBI 502 Database Management Systems Example 2 Who has a loan with a greater value than every loan at Perryridge branch? t1 ←  mp(max) ( g max(amount) (  branch-name = “Perryridge” (loan))) (  customer-name (  loan-number (  amount > t1.max (loan)) ⋈ borrower)

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.12 of 48 UBI 502 Database Management Systems Example 2 (cont)  branch-name = “Perryridge” (loan) mysql> SELECT * FROM loan -> WHERE branch_name = "Perryridge"; | loan_number | branch_name | amount | | L-15 | Perryridge | 1500 | | L-16 | Perryridge | 1300 |

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.13 of 48 UBI 502 Database Management Systems Example 2 (cont) g max(amount) (  branch-name = “Perryridge” (loan)) mysql> SELECT MAX(amount) FROM loan -> WHERE branch_name = "Perryridge"; | max(amount) | | 1500 |

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.14 of 48 UBI 502 Database Management Systems Example 2 (cont)  mp(max) ( g max(amount) (  branch-name = “Perryridge” (loan))) mysql> SELECT MAX(amount) AS max FROM loan -> WHERE branch_name = "Perryridge"; | max | | 1500 |

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.15 of 48 UBI 502 Database Management Systems Example 2 (cont) t1 ←  mp(max) ( g max(amount) (  branch-name = “Perryridge” (loan))) mysql> CREATE TEMPORARY TABLE t1 -> SELECT MAX(amount) AS max FROM loan -> WHERE branch_name = "Perryridge";

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.16 of 48 UBI 502 Database Management Systems Example 2 (cont)  amount > t1.max (loan) mysql> SELECT * FROM loan, t1 -> WHERE amount > t1.max; | loan_number | branch_name | amount | max | | L-23 | Redwood | 2000 | 1500 | | L-20 | North Town | 7500 | 1500 |

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.17 of 48 UBI 502 Database Management Systems Example 2 (cont) (  amount (  amount > t1.max (loan)) mysql> SELECT amount FROM loan, t1 -> WHERE amount > t1.max; | amount | | 2000 | | 7500 |

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.18 of 48 UBI 502 Database Management Systems Example 2 (cont)  customer-name (  amount (  amount > t1.max (loan)) ⋈ borrower) mysql> SELECT customer_name -> FROM loan NATURAL JOIN borrower, t1 -> WHERE amount > t1.max; | customer_name | | McBride | | Smith |

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.19 of 48 UBI 502 Database Management Systems Example 3 Query: What is the average loan amount held by customers in each city? branch-city g avg(amount) (loan ⋈ branch)

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.20 of 48 UBI 502 Database Management Systems Example 3 (cont) mysql> select * -> from loan natural join branch; | loan_number | branch_name | amount | branch_name | branch_city | assets | | L-17 | Downtown | 1000 | Downtown | Brooklyn | | | L-23 | Redwood | 2000 | Redwood | Palo Alto | | | L-15 | Perryridge | 1500 | Perryridge | Horseneck | | | L-14 | Downtown | 1500 | Downtown | Brooklyn | | | L-93 | Mianus | 500 | Mianus | Horseneck | | | L-11 | Round Hill | 900 | Round Hill | Horseneck | | | L-16 | Perryridge | 1300 | Perryridge | Horseneck | | | L-20 | North Town | 7500 | North Town | Rye | | | L-21 | Central | 570 | Central | Rye | | Would like to see just the branch_name and branch_city columns

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.21 of 48 UBI 502 Database Management Systems Example 3 (cont) mysql> select branch_city, amount -> from loan natural join branch; | branch_city | amount | | Brooklyn | 1000 | | Palo Alto | 2000 | | Horseneck | 1500 | | Brooklyn | 1500 | | Horseneck | 500 | | Horseneck | 900 | | Horseneck | 1300 | | Rye | 7500 | | Rye | 570 | Would like to see this table sorted by branch_city

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.22 of 48 UBI 502 Database Management Systems Example 3 (cont) mysql> select branch_city, amount -> from loan natural join branch -> order by branch_city; | branch_city | amount | | Brooklyn | 1000 | | Brooklyn | 1500 | | Horseneck | 1300 | | Horseneck | 1500 | | Horseneck | 500 | | Horseneck | 900 | | Palo Alto | 2000 | | Rye | 7500 | | Rye | 570 | Next: compute average for each group

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.23 of 48 UBI 502 Database Management Systems Example 3 (cont) mysql> select branch_city, avg(amount) -> from loan natural join branch -> group by branch_city; | branch_city | avg(amount) | | Brooklyn | | | Horseneck | | | Palo Alto | | | Rye | | Suppose we wanted to impose a constraint on the result – use a HAVING clause

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.24 of 48 UBI 502 Database Management Systems Example 3 (cont) mysql> select branch_city as city, avg(amount) as average -> from loan natural join branch -> group by branch_city -> having (avg(amount) > 1500); | city | average | | Palo Alto | | | Rye | | Note: we also renamed the columns

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.25 of 48 UBI 502 Database Management Systems Join tricks Restrict a to just those rows compatible with b  SELECT DISTINCT a.* FROM a NATURAL JOIN b; E.g. list tuples from the borrower table only for those borrowers who are depositors:  SELECT DISTINCT borrower.* FROM borrower NATURAL JOIN depositor;

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.26 of 48 UBI 502 Database Management Systems Join tricks (cont) Compute set difference without using EXCEPT (MySQL 3.23 doesn’t have set operations) i.e.remove from a those tuples compatible with b  SELECT DISTINCT a.* FROM a NATURAL LEFT OUTER JOIN b WHERE b.x is NULL E.g. list those tuples from the borrower table only for those borrowers who are not depositors:  SELECT DISTINCT borrower.* FROM borrower NATURAL LEFT OUTER JOIN depositor WHERE depositor.customer_name is NULLS

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.27 of 48 UBI 502 Database Management Systems Review A NATURAL JOIN uses common column headings, and it is not possible to attach an ON or USING clause An INNER JOIN results in a table where each tuple is a combination of information from both argument tables An OUTER JOIN is an INNER JOIN padded with additional tuples The ON clause lists common attributes which must be equal on both sides of the join The USING clause permits us to enforce equality between attributes having different names, e.g. ON a.x = b.y It also permits us to use inequalities, e.g. ON a.x > b.y

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.28 of 48 UBI 502 Database Management Systems PS: Theta Join A ⋈ C B ≝  C (A ⋈ B) For notational convenience, we permit ourselves to write a select condition on the join operator This is called “theta join” because the condition is sometimes represented as a theta subscript: ⋈ θ

©Silberschatz, Korth and Sudarshan Modifications & additions by Cengiz Güngör 5.29 of 48 UBI 502 Database Management Systems READING MySQL manual chapter 3: Tutorial Introduction bychapter/manual_Tutorial.html