group by and select max with value null - sql

Question

I have a next problem with query
SELECT
T.DETALLE_BECA_ANIO anio,
T.DETALLE_BECA_MES mes,
T.DETALLE_BECA_NIVEL_EDU_ID edu_id,
T.DETALLE_BECA_TRAMO_ID tr_id,
MAX(
CASE
WHEN T.DETALLE_BECA_TIPO_BENE_ID IS NULL
THEN NVL(DETALLE_BECA_VALOR,0)
ELSE 0
END) mant ,
MAX(
CASE
WHEN T.DETALLE_BECA_TIPO_BENE_ID = 1
THEN NVL(DETALLE_BECA_VALOR,0)
ELSE 0
END) tras
FROM
(SELECT DETALLE_BECA_NIVEL_EDU_ID,
DETALLE_BECA_BECA_ID,
DETALLE_BECA_TIPO_BENE_ID,
DETALLE_BECA_VALOR,
DETALLE_BECA_MES,
DETALLE_BECA_REGION_ID,
DETALLE_BECA_PROVINCIA_ID,
DETALLE_BECA_ANIO,
DETALLE_BECA_TRAMO_ID,
DETALLE_BECA_COMUNA_ID
FROM TBL_DETALLE_BECAS
WHERE (DETALLE_BECA_TIPO_BENE_ID = 1
OR DETALLE_BECA_TIPO_BENE_ID IS NULL)
and DETALLE_BECA_BECA_ID = 1
and detalle_beca_mes = 3
) T
GROUP BY T.DETALLE_BECA_BECA_ID,
T.DETALLE_BECA_TRAMO_ID,
T.DETALLE_BECA_REGION_ID,
T.DETALLE_BECA_PROVINCIA_ID,
T.DETALLE_BECA_ANIO,
T.DETALLE_BECA_MES,
T.DETALLE_BECA_NIVEL_EDU_ID,
T.DETALLE_BECA_COMUNA_ID
ORDER BY T.DETALLE_BECA_BECA_ID,
T.DETALLE_BECA_MES,
T.DETALLE_BECA_NIVEL_EDU_ID
output:
"ANIO" "MES" "EDU_ID" "TR_ID" "MANT" "TRAS"
2017 3 2 0.62 0 NULL
2017 3 3 1.24 6 NULL
2017 3 NULL 1.0 NULL 1
I need that sum value where EDU_ID is null with value 2,3 in TR_ID and replace value null in "tras" with value from EDU is null
"ANIO" "MES" "EDU_ID" "TR_ID" "MANT" "TRAS"
2017 3 2 1.62 0 1
2017 3 3 2.24 6 1
I writed query with min(edu_id) or max(edu_id ) but could not solve my problem.
The other thing that occurred to me is to make a join with the same table

Answer 1

First, this makes more sense as your query:
SELECT T.DETALLE_BECA_ANIO as anio, T.DETALLE_BECA_MES as mes,
T.DETALLE_BECA_NIVEL_EDU_ID as edu_id, T.DETALLE_BECA_TRAMO_ID as tr_id,
MAX(CASE WHEN T.DETALLE_BECA_TIPO_BENE_ID IS NULL
THEN NVL(DETALLE_BECA_VALOR, 0)
ELSE 0
END) as mant ,
MAX(CASE WHEN T.DETALLE_BECA_TIPO_BENE_ID = 1
THEN NVL(DETALLE_BECA_VALOR,0)
ELSE 0
END) tras
FROM TBL_DETALLE_BECAS
WHERE (DETALLE_BECA_TIPO_BENE_ID = 1 OR DETALLE_BECA_TIPO_BENE_ID IS NULL) AND
DETALLE_BECA_BECA_ID = 1 AND
detalle_beca_mes = 3
GROUP BY T.DETALLE_BECA_ANIO, T.DETALLE_BECA_MES,
T.DETALLE_BECA_NIVEL_EDU_ID, T.DETALLE_BECA_TRAMO_ID
ORDER BY T.DETALLE_BECA_BECA_ID, T.DETALLE_BECA_MES, T.DETALLE_BECA_NIVEL_EDU_ID;
This eliminates the subquery (unnecessary) and only aggregates by the columns being returned. A proper query might fix your problem.
But, you seem to want to use NULL to be "all" for the other columns. If so, something like this will work:
WITH t as (
SELECT T.DETALLE_BECA_ANIO as anio, T.DETALLE_BECA_MES as mes,
T.DETALLE_BECA_NIVEL_EDU_ID as edu_id, T.DETALLE_BECA_TRAMO_ID as tr_id,
MAX(CASE WHEN T.DETALLE_BECA_TIPO_BENE_ID IS NULL
THEN NVL(DETALLE_BECA_VALOR, 0)
ELSE 0
END) as mant ,
MAX(CASE WHEN T.DETALLE_BECA_TIPO_BENE_ID = 1
THEN NVL(DETALLE_BECA_VALOR,0)
ELSE 0
END) tras
FROM TBL_DETALLE_BECAS
WHERE (DETALLE_BECA_TIPO_BENE_ID = 1 OR DETALLE_BECA_TIPO_BENE_ID IS NULL) AND
DETALLE_BECA_BECA_ID = 1 AND
detalle_beca_mes = 3
GROUP BY T.DETALLE_BECA_ANIO, T.DETALLE_BECA_MES,
T.DETALLE_BECA_NIVEL_EDU_ID, T.DETALLE_BECA_TRAMO_ID
)
SELECT t.ANIO, t.MES, t.EDU_ID,
COALESCE(t.TR_ID, 0) + COALESCE(tnull.TR_ID, 0) as TR_ID,
t.MANT,
COALESCE(t.TRAS, 0) + COALESCE(tnull.TRAS, 0) as TRAS
FROM t LEFT JOIN
(SELECT t.*
FROM t
WHERE t.edu_id IS NULL
) tnull
ON tnull.ANIO = t.ANIO AND tnull.MES = t.MES
WHERE t.edu_id IS NOT NULL
ORDER BY T.DETALLE_BECA_BECA_ID, T.DETALLE_BECA_MES, T.DETALLE_BECA_NIVEL_EDU_ID;