INSERT aggregated values from a query into a table in SQL - postgresql

Question

I'm new to SQL and proceeded much by trial and error as well as searching books and the internet. I have to repeat a query for the sum over monthly data for five years and I'd like to insert the results for every month as a column in a table. I tried adding new columns for every month
alter table add column, insert etc.
but I can't get it right. Here's the code I used for jan and feb07:
CREATE TABLE "TVD_db"."lebendetiere"
(nuar text,
ak text,
sex text,
jan07 text,
feb07 text,
märz07 text,
april07 text,
mai07 text,
juni07 text,
juli07 text,
aug07 text,
sept07 text,
okt07 text,
nov07 text,
dez07 text,
jan08 text,
....
dez11 text);
INSERT INTO "TVD_db"."lebendetiere" (nuar, ak, sex, jan07)
SELECT
"AUFENTHALTE"."nuar",
CASE WHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") < 365 THEN '1' WHEN DATE('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") > 730 THEN 3 ELSE 2 END AS AK,
CASE WHEN "AUFENTHALTE"."isweiblich" = 'T' THEN 'female' ELSE 'male' END AS sex,
COUNT("AUFENTHALTE"."tierid")
FROM "TVD_db"."AUFENTHALTE"
WHERE DATE("AUFENTHALTE"."gueltigvon") <= DATE('2007-01-01')
AND DATE("AUFENTHALTE"."gueltigbis") >= DATE('2007-01-01')
GROUP BY "AUFENTHALTE"."nuar",
CASE WHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") < 365 THEN '1' WHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") > 730 THEN 3 ELSE 2 END,
CASE WHEN "AUFENTHALTE"."isweiblich" = 'T' THEN 'female' ELSE 'male' END
ORDER BY "AUFENTHALTE"."nuar",
CASE WHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") < 365 THEN '1' wWHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") > 730 THEN 3 ELSE 2 END,
CASE WHEN "AUFENTHALTE"."isweiblich" = 'T' THEN 'female' ELSE 'male' END
;
--until here it works fine
UPDATE "TVD_db"."lebendetiere" SET "feb07"= --this is the part I cant get right...
(SELECT
COUNT("AUFENTHALTE"."tierid")
FROM "TVD_db"."AUFENTHALTE"
WHERE DATE("AUFENTHALTE"."gueltigvon") <= DATE('2007-02-01')
AND DATE("AUFENTHALTE"."gueltigbis") >= DATE('2007-02-01')
GROUP BY "AUFENTHALTE"."nuar",
CASE WHEN DATE ('2007-02-01')- DATE ("AUFENTHALTE"."gebdat") < 365 THEN '1' WHEN DATE ('2007-02-01')- DATE ("AUFENTHALTE"."gebdat") > 730 THEN 3 ELSE 2 END,
CASE WHEN "AUFENTHALTE"."isweiblich" = 'T' THEN 'female' ELSE 'male' END
ORDER BY "AUFENTHALTE"."nuar",
CASE WHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") < 365 THEN '1' wWHEN DATE ('2007-01-01')- DATE ("AUFENTHALTE"."gebdat") > 730 THEN 3 ELSE 2 END,
CASE WHEN "AUFENTHALTE"."isweiblich" = 'T' THEN 'female' ELSE 'male' END);
Has anyone a solution or do I have to make a table for every month and then join the results?

Answer 1

After reading your post thoroughly, here is a complete redesign that should hold some insight for beginners in the field of SQL / PostgreSQL.
I would advise not to use mixed case identifiers in PostgreSQL. Use lower case exclusively, then you don't have to double-quote them and your code is much easier to read. You also avoid a lot of possible confusion.
Use table aliases to make your code more readable.
Column names in the SELECT statement for the INSERT are irrelevant. That's why I commented then out (avoids possible naming conflicts).
Use ordinal numbers in GROUP BY and ORDER BY to further simplify.
Don't use a separate column for every new month. Use a column identifying the month and add a row per month.
If you actually need the design with one column per month, then you need a large CASE statement or a pivot query. Refer to the tablefunc extension. But this is complicated stuff for an SQL newbie. I really think, you want a row per month.
I use generate_series() to generate one row per month between Jan 2007 and Dec 2011.
With my changed design, you don't need extra UPDATEs. It's all done in one INSERT.
I simplified quite a couple of other things. Here is what I would propose instead:
CREATE TABLE tvd_db.lebendetiere(
nuar text,
,alterskat integer
,sex text
,datum date
,anzahl integer
);
INSERT INTO tvd_db.lebendetiere (nuar, alterskat, sex, datum, anzahl)
SELECT a.nuar
,CASE WHEN a.gebdat >= '2006-01-01'::date THEN 1 -- use >= !
WHEN a.gebdat < '2005-01-01'::date THEN 3
ELSE 2 END -- AS alterskat
,CASE WHEN a.isweiblich = 'T' THEN 'female' ELSE 'male' END -- AS sex
,m.m
,count(*) -- AS anzahl
FROM tvd_db.aufenthalte a
CROSS JOIN (
SELECT generate_series('2007-01-01'::date
,'2011-12-01'::date, interval '1 month')::date
) m(m)
WHERE a.gueltigvon <= m.m
AND a.gueltigbis >= m.m
GROUP BY a.nuar, 2, 3, m.m
ORDER BY a.nuar, 2, 3, m.m;