I have a column in an SQL table. That contains a dateof delivery order.
So the same date can be repeat (in one day we delivred severals orders), like this:
05-01-16 05-01-16 05-01-16 08-01-16 08-01-16 14-01-16 22-01-16 22-01-16 04-02-16 05-02-16 05-02-16 I want, compute the AVG of the turnover of each article in each 6 months, I explain more:
From January to June ==> Turnover 1 From Febrary to July ==> Turnover 2 From March to August ==> Turnover 3 From April to September ==> Turnover 4 From May to Obtober ==> Turnover 5 From June to November ==> Turnover 6 From July to December ==> Turnover 7 I'm already extracted the month by the request bellow, but I can't compute dynamically (because my data should be change each month) the turnover like this example above:
select distinct extract (month from Article) as mt order by mt I tried to use a cursor but I can't arrived to the best solution.
I did a request to compute a turnover for each customer per article in the first 6 months (I did it manually ) is the following:
select "LRU", "Client", round(sum("Montant_fac_eur")) from "foundry" where "Nature"='Repair' and "Client"={{w_widget3.selectedValue}} and "annee"='2016' and extract (month from "date") between '1' and '6' group by "LRU", "Client" Her result is the following:
LRU Client round "article1" 4001 8859 Turnover of article1 from January to June "article2" 4001 94315 Turnover of article2 from January to June "article3" 4001 273487 Turnover of article3 from January to June "article4" 4001 22292 Turnover of article4 from January to June "article5" 4001 22292 Turnover of article5 from January to June "article6" 4001 42590 Turnover of article6 from January to June "article7" 4001 9965 Turnover of article7 from January to June "article8" 4001 39654 Turnover of article8 from January to June "article9" 4001 3883 Turnover of article9 from January to June "article10" 4001 41612 Turnover of article10 from January to June I want do a loop to compute a turnover each 6 months without to write it manually if it possible ? Can someones please help me and give me a solution or suggestion how can I do it ? Thank you.
4 Answers
Answers 1
Here you can view the simplified definition and solution of your problem (if I understood you correctly): http://sqlfiddle.com/#!9/48a2e1/1
CREATE TABLE foundry ( lru varchar(50) NOT NULL, client int NOT NULL, purchase_date date, price int NOT NULL ); INSERT INTO foundry (lru, client, purchase_date, price) VALUES ("article1", 4001, "01-01-16", 100), ("article1", 4001, "01-01-17", 200), ("article1", 4001, "01-02-16", 300), ("article1", 4001, "01-04-16", 400), ("article1", 4001, "01-06-16", 500), ("article1", 4001, "01-08-16", 600), ("article1", 4001, "01-10-16", 700), ("article1", 4001, "01-11-16", 800), ("article1", 4002, "01-01-16", 900), ("article1", 4002, "01-07-16", 1000), ("article1", 4002, "01-12-16", 1100); Basically we have a table with four columns: lru (article name), client, purchase date, and some price.
The solution looks like:
SELECT lru, client, avg(price), COUNT(*) as total_items, MONTHNAME(STR_TO_DATE(L, '%m')) as start_month, MONTHNAME(STR_TO_DATE(R, '%m')) as end_month FROM foundry, ( SELECT 1 as L, 6 as R UNION ALL SELECT 2, 7 UNION ALL SELECT 3, 8 UNION ALL SELECT 4, 9 UNION ALL SELECT 5, 10 UNION ALL SELECT 6, 11 UNION ALL SELECT 7, 12 ) months WHERE month(purchase_date) >= L AND month(purchase_date) <= R GROUP BY lru, client, L, R The idea is:
- Generate all possible combinations of months: 1-6, 2-7, ..., 7,12
- Join source data with generated months combination
- Use AVG with GROUP BY
And the result:
lru client avg(price) total_items start_month end_month article1 4001 300 5 January June article1 4001 400 3 February July article1 4001 500 3 March August article1 4001 500 3 April September article1 4001 600 3 May October article1 4001 650 4 June November article1 4001 700 3 July December article1 4002 900 1 January June article1 4002 1000 1 February July article1 4002 1000 1 March August article1 4002 1000 1 April September article1 4002 1000 1 May October article1 4002 1000 1 June November article1 4002 1050 2 July December Answers 2
I would recommend common table expressions for each six-month coverage. https://technet.microsoft.com/en-us/library/ms186243(v=sql.105).aspx Like: with tp1 (Turnoverperiod, Averagevaule) as ( select 'Period1' as Turnoverperiod, AVG(turnover) as Averagevalue where date between period1.startdate and period2.enddate ) , tp2 as ( .... tp2 )
select * from tp1 union select * from tp2
also, you can create a dynamic sql string (nvarchar(max)) you can programatically append the union queries then use sp_executesql statement.
Answers 3
I am not sure which RDBMs you are using, so here a possible solution with ANSI SQL (use your RDBMs analogue)
Try address this problem in 2 steps:
Create a view in a form (lets say you call it v_turnover_per_month), using basic group by syntax over your source table:
article month turnover art1 201701 1000 art1 201702 1020 ... ... ... art2 201701 5000 ... ... ...Use following select statement
SELECT article, 'Turnover from ' || m1.month || ' until ' || m6.month as title, max(m1.turnover + m2.turnover + m3.turnover + m4.turnover + m5.turnover + m6.turnover) as total_turnover_6month FROM v_turnover_per_month as m6 JOIN v_turnover_per_month as m5 ON m6.article = m5.article and m5.month+1=m6.month JOIN v_turnover_per_month as m4 ON m6.article = m4.article and m4.month+1=m5.month JOIN v_turnover_per_month as m3 ON m6.article = m3.article and m3.month+1=m4.month JOIN v_turnover_per_month as m2 ON m6.article = m2.article and m2.month+1=m3.month JOIN v_turnover_per_month as m1 ON m6.article = m1.article and m1.month+1=m2.month GROUP BY 1, 2;
Answers 4
Here's my take:
DROP TABLE IF EXISTS test; # credit to tvelykyy for providing some test data and table def which I adapted CREATE TABLE test (lru VARCHAR(16), `client` INTEGER UNSIGNED, purchase_date DATE, price int NOT NULL); INSERT INTO test (lru, client, purchase_date, price) VALUES ("article1", 4001, '2016-01-01', 100), ("article1", 4001, '2016-02-01', 200), ("article1", 4001, '2016-03-01', 300), ("article1", 4001, '2016-04-01', 400), ("article1", 4001, '2016-05-01', 500), ("article1", 4001, '2016-06-01', 600), ("article1", 4001, '2016-07-01', 700), ("article1", 4001, '2016-08-01', 800), ("article1", 4002, '2016-01-01', 1100), ("article1", 4002, '2016-06-01', 1200); SELECT A.lru, A.`client`, ROUND(SUM(price)) round, CONCAT('Turnover of article ', A.lru, ' from ', DATE_FORMAT(DATE_ADD(MAX(B.purchase_date), INTERVAL -6 MONTH), '%M %Y'), ' to ', DATE_FORMAT(MAX(B.purchase_date), '%M %Y')) period FROM ( SELECT DISTINCT lru, LAST_DAY(purchase_date) purchase_month, `client` FROM test ) A LEFT OUTER JOIN test B ON A.lru = B.lru AND A.`client` = B.`client` AND A.purchase_month >= LAST_DAY(B.purchase_date) AND DATE_ADD(A.purchase_month, INTERVAL -6 MONTH) < LAST_DAY(B.purchase_date) GROUP BY A.lru, A.`client`, A.purchase_month ORDER BY A.lru, A.`client`, A.purchase_month; The query works by selecting the DISTINCT lru, clients and months in the subquery A which allows a LEFT JOIN back to the original data ON the lru, client and date in the last six months using the last day of each month to define the months in a standard manner. Finally aggregation is used as specified.
Please let me know if you require further information or I didn't understand the question in its entirety.
Thanks,
James
0 comments:
Post a Comment