CODEWITHSUNDEEP
sqldb2
Ib D
Ib D

Reputation: 441

Group By combined to CASE (SQL DB2)

I have a database of customers with this columns (sample) :

ID TIMESTAMP TYP_EVEN NAME COUNTRY GENDER NBR_CHILDREN ADRESS CUSTOMER_LINKED TYP_OF_LINK
044348547 2020-09-08-02.02.21.442908 0004 NXXX CORINNE FRA 2 02 000000000
044379039 2020-07-17-11.17.55.410843 0013 00 11 RUE XXXX XXX 000000000
044379039 2020-07-21-16.45.53.485200 0004 KXXX STEPHANE FRA 1 00 000000000
044379039 2020-08-05-02.02.41.403053 0004 KXXX STEPHANE FRA 1 00 000000000
044385911 2019-01-31-02.03.22.455660 0004 SXXX PHILIPPE FRA 1 02 000000000
044385911 2020-09-18-15.46.31.387588 0004 SXXX PHILIPPE FRA 1 02 000000000
044385911 2020-09-30-02.01.45.310449 0004 SXXX PHILIPPE FRA 1 02 000000000
044385911 2020-10-01-00.06.20.341427 0013 00 19 RUE XXXX XXXX 000000000
044680501 2021-03-10-02.02.42.979297 0004 HXXX STEPHANE FRA 1 00 000000000
044680501 2021-03-10-02.02.42.980069 0013 00 22 RUE XXXX XXXX 000000000
044680501 2021-03-10-02.02.42.981083 0029 00 044680519 MARIED
044680519 2021-03-10-02.02.42.966522 0004 HXXX LIDWINE FRA 2 02 000000000
044680519 2021-03-10-02.02.42.970031 0029 00 044680501 MARIED

TYP_EVEN : is type of event about the record for each customers ID, then OOO4 : is about informations of the customer (NAME, COUNTRY, GENDER, Number of children), 0013 : is about the ADRESS informations and 0029 : is about link beetwen customers

I would like to have one row by ID with all informations. I proceed like this :

    SELECT
T1.ID,
CASE WHEN T1.TYP_EVEN = '0004' THEN T1.NAME END AS NAME,
CASE WHEN T1.TYP_EVEN = '0004' THEN T1.COUNTRY END AS COUNTRY,
CASE WHEN T1.TYP_EVEN = '0004' THEN T1.GENDER END AS GENDER,
CASE WHEN T1.TYP_EVEN = '0004' THEN T1.NBR_CHILDREN END AS NBR_CHILDREN,

CASE WHEN T1.TYP_EVEN = '0013' THEN T1.ADRESS END AS ADRESS,

CASE WHEN T1.TYP_EVEN = '0029' THEN T1.CUSTOMER_LINKED END AS CUSTOMER_LINKED,
CASE WHEN T1.TYP_EVEN = '0029' THEN T1.TYP_OF_LINK END AS TYP_OF_LINK
,T1.TIMESTAMP

FROM DB.CUSTOMERS T1

GROUP BY GROUP BY T1.ID,T1.TIMESTAMP, T1.TYP_EVEN,T1.NAME,T1.COUNTRY,T1.GENDER,T1.NBR_CHILDREN,T1.ADRESS
,T1.CUSTOMER_LINKED, T1.TYP_OF_LINK 
ORDER BY 1

the result should be :

ID NAME COUNTRY GENDER NBR_CHILDREN ADDRESS CUSTOMER_LINKED TYP_OF_LINK
044348547 NXXX CORINNE FRA 2 02 11 RUE XXXX XXX 000000000
044379039 KXXX STEPHANE FRA 1 00 000000000
044385911 SXXX PHILIPPE FRA 1 02 19 RUE XXXX XXXX 000000000
044680501 HXXX STEPHANE FRA 1 00 22 RUE XXXX XXXX 044680519 MARIED
044680519 HXXX LIDWINE FRA 2 02 044680501 MARIED

More information: if an ID have 2 rows with same TYP_EVEN then choose the latter one by the TIMESTAMP.

Could someone have any idea to resolve this SQL Statement ?

Upvotes: 0

Views: 108

Answers (1)

nfgl
nfgl

Reputation: 3202

Here is a way to do what you want

with customers (id, timestamp, typ_even, name, country, gender, nbr_children, adress, customer_linked, typ_of_link) as (
  values
  (044348547, timestamp('2020-09-08-02.02.21.442908'), 0004, 'NXXX CORINNE', 'FRA', 2, 02, NULL, 000000000, NULL),
  (044379039, '2020-07-17-11.17.55.410843', 0013, NULL, NULL, NULL, 00, '11 RUE XXXX XXX', 000000000, NULL),
  (044379039, '2020-07-21-16.45.53.485200', 0004, 'KXXX STEPHANE', 'FRA', 1, 00, NULL, 000000000, NULL),
  (044379039, '2020-08-05-02.02.41.403053', 0004, 'KXXX STEPHANE', 'FRA', 1, 00, NULL, 000000000, NULL),
  (044385911, '2019-01-31-02.03.22.455660', 0004, 'SXXX PHILIPPE', 'FRA', 1, 02, NULL, 000000000, NULL),
  (044385911, '2020-09-18-15.46.31.387588', 0004, 'SXXX PHILIPPE', 'FRA', 1, 02, NULL, 000000000, NULL),
  (044385911, '2020-09-30-02.01.45.310449', 0004, 'SXXX PHILIPPE', 'FRA', 1, 02, NULL, 000000000, NULL),
  (044385911, '2020-10-01-00.06.20.341427', 0013, NULL, NULL, NULL, 00, '19 RUE XXXX XXXX', 000000000, NULL),
  (044680501, '2021-03-10-02.02.42.979297', 0004, 'HXXX STEPHANE', 'FRA', 1, 00, NULL, 000000000, NULL),
  (044680501, '2021-03-10-02.02.42.980069', 0013, NULL, NULL, NULL, 00, '22 RUE XXXX XXXX', 000000000, NULL),
  (044680501, '2021-03-10-02.02.42.981083', 0029, NULL, NULL, NULL, 00, NULL, 044680519, 'MARIED'),
  (044680519, '2021-03-10-02.02.42.966522', 0004, 'HXXX LIDWINE', 'FRA', 2, 02, NULL, 000000000, NULL),
  (044680519, '2021-03-10-02.02.42.970031', 0029, NULL, NULL, NULL, 00, NULL, 044680501, 'MARIED')
)
select ID,NAME,COUNTRY,GENDER,NBR_CHILDREN,ADRESS,CUSTOMER_LINKED,TYP_OF_LINK from (
  select
    id,
    last_value(name, 'IGNORE NULLS') over(partition by id order by timestamp) name,
    last_value(country, 'IGNORE NULLS') over(partition by id order by timestamp) country,
    last_value(gender, 'IGNORE NULLS') over(partition by id order by timestamp) gender,
    last_value(nbr_children, 'IGNORE NULLS') over(partition by id order by timestamp) nbr_children,
    last_value(adress, 'IGNORE NULLS') over(partition by id order by timestamp ) adress,
    last_value(customer_linked, 'IGNORE NULLS') over(partition by id order by timestamp ) customer_linked,
    last_value(typ_of_link, 'IGNORE NULLS') over(partition by id order by timestamp ) typ_of_link,
    rownumber() over(partition by id order by timestamp desc) seq
  from customers
) where seq = 1
order by id

See db<>fiddle, Corinne NXXX has no address

Upvotes: 1

Related Questions