This lecture. Databases - SQL II. Counting students. Summary Functions

Transcription

1 This lecture Databases - SQL II This lecture focuses on the summary or aggregate features provided in MySQL. The summary functions are those functions that return a single value from a collection of values for example, functions that produce counts, totals and averages. (GF Royle , N Spadaccini 2008) Structured Query Language II 1 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 2 / 22 Summary Functions Counting students One of the main uses of a database is to summarize the data it contains, in particular to provide statistical data. The main summary functions are COUNT to count rows SUM to add the values in a column MIN to find the minimum value in a column MAX to find the maximum value in a column AVG to find the average value in a column STD to find the standard deviation of the values in a column How many students are in the class for the grade-keeping project? SELECT COUNT(*) FROM student; 31 The COUNT function says to count the number of rows that are returned by the SELECT statement notice that this syntax is not very intuitive. (GF Royle , N Spadaccini 2008) Structured Query Language II 3 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 4 / 22

2 How many men and women? If we add a WHERE clause to the statement, then the COUNT will apply only to the selected rows. SELECT COUNT(*) FROM student WHERE sex = M ; 16 SELECT COUNT(*) FROM student WHERE sex = F ; 15 With one statement We can count both men and women in a single statement by using the GROUP BY clause. SELECT COUNT(*) FROM student GROUP BY sex; (GF Royle , N Spadaccini 2008) Structured Query Language II 5 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 6 / 22 But which is which Statistical Data As it stands, we don t know which value is associated with which sex! SELECT sex, COUNT(*) FROM student GROUP BY sex; sex F 15 M The GROUP BY clause says to first group the rows according to the distinct values of the specified attribute(s) and then do the counting. Now let s try and find statistical data about the quizzes and tests. SELECT event_id, MIN(score), MAX(score), AVG(score) FROM score GROUP BY event_id; event_id MIN(score) MAX(score) AVG(score) (GF Royle , N Spadaccini 2008) Structured Query Language II 7 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 8 / 22

3 Counting tests and quizzes Separating tests and quizzes How many of the events were tests and how many were quizzes? SELECT G.category, COUNT(*) FROM grade_event G GROUP BY G.category; category T 2 Q Can we get separate summary data for the quizzes and the tests? To do this we will need to do a multi-table query because score does not know what type each event is. SELECT G.category, AVG(S.score) FROM grade_event G, score S WHERE G.event_id = S.event_id GROUP BY G.category; category AVG(S.score) T Q (GF Royle , N Spadaccini 2008) Structured Query Language II 9 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 10 / 22 Separating males and females Super-aggregate Now suppose we want to find the averages for each sex separately and for tests and quizzes separately. SELECT G.category, S.sex, AVG(M.score) WHERE G.event_id = M.event_id AND M.student_id = S.student_id GROUP BY G.category, S.sex; category sex AVG(M.score) T F T M Q F Q M SELECT G.category, S.sex, AVG(M.score) WHERE G.event_id = M.event_id AND M.student_id = S.student_id GROUP BY G.category, S.sex WITH ROLLUP; category sex AVG(M.score) Q F Q M Q NULL T F T M T NULL NULL NULL (GF Royle , N Spadaccini 2008) Structured Query Language II 11 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 12 / 22

4 What rollup does Adding the names The ROLLUP clause generates summaries of summaries that are inserted at appropriate places in the table. The GROUP BY clauses caused the data to summarised according to the four groups (Q, F), (Q, M), (T, F), (T, M). Rollup causes these groups to be further grouped together into (Q, both) and (T, both) and then finally combined into a single group. The fields where multiple values have been counted together are displayed in the result set by using NULL for that field. At the end of semester, the lecturer needs to know how many marks each person got in their quizzes and tests. SELECT S.name, G.category, COUNT(*), SUM(M.score) WHERE G.event_id = M.event_id AND S.student_id = M.student_id GROUP BY S.name, G.category WITH ROLLUP; (GF Royle , N Spadaccini 2008) Structured Query Language II 13 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 14 / 22 The output name category SUM(M.score) Abby Q 4 63 Abby T Abby NULL Aubrey Q 4 58 Aubrey T Aubrey NULL Avery Q 3 40 Avery T Avery NULL Becca Q 4 60 Becca T Filtering on aggregate values Suppose we want to find the student who got the highest average quiz mark. SELECT S.name, COUNT(*), AVG(M.score) WHERE G.category = Q AND G.event_id = M.event_id AND S.student_id = M.student_id GROUP BY S.name ORDER BY AVG(M.score) DESC; name AVG(M.score) Megan Gabrielle Michael Teddy (GF Royle , N Spadaccini 2008) Structured Query Language II 15 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 16 / 22

5 Using HAVING But the quiz-prize can only go to a student who sat all of the quizzes. SELECT S.name, COUNT(*), AVG(M.score) WHERE G.category = Q AND G.event_id = M.event_id AND S.student_id = M.student_id GROUP BY S.name HAVING COUNT(*) = 4 ORDER BY AVG(M.score) DESC; name AVG(M.score) Michael Teddy Summary The HAVING clause behaves exactly like a WHERE clause except that it operates on the summarized data, so the whole process is as follows: The named columns are extracted from the Cartesian product of all the tables listed in the FROM clause. All of these rows are then filtered according to the WHERE clause The filtered rows are then grouped together according to the GROUP BY clause The aggregate functions are applied to the rows in each group. The resulting rows are then filtered by the HAVING clause. The filtered, aggregated rows are then ordered by the ORDER BY clause. (GF Royle , N Spadaccini 2008) Structured Query Language II 17 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 18 / 22 Using DISTINCT In order to count the number of different states from which the presidents come, we can use SELECT COUNT (DISTINCT state) FROM president; COUNT(DISTINCT state) The DISTINCT keyword eliminates the duplicate values before counting. Tables with NULL values Consider a table with the following data mysql> select * from test; mark NULL What is the number of rows, the sum of the rows and the average value for the single field? (GF Royle , N Spadaccini 2008) Structured Query Language II 19 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 20 / 22

6 Sometimes NULL counts, sometimes not! Learning how to summarize mysql> SELECT COUNT(*), SUM(mark), AVG(mark) FROM test; SUM(mark) AVG(mark) Notice that AVG is not equal to SUM / COUNT. Learning how to use the summary functions requires a lot of practice because you just have to learn the somewhat strange syntax, and the error messages produced by MySQL are not very informative. For example, to someone used to a normal programming language it seems very strange to type SELECT COUNT(*) FROM president rather than WHERE death IS NULL; COUNT (SELECT * FROM president WHERE death is NULL); (GF Royle , N Spadaccini 2008) Structured Query Language II 21 / 22 (GF Royle , N Spadaccini 2008) Structured Query Language II 22 / 22