Case statement COUNT for THEN in (PostgreSQL) 9.3.11 - sql

Question

Firstly, this is a coursework question, so I am not looking for a full answer, just a hint :)
I have a "monarch" database that stores name, house(?), accession, coronation(?) that keeps track of monarchs (including prime ministers). The house and coronation applies only to monarchs and returns null if monarch is a prime minister.
It looks as follows:
I am required to write a psql query that returns the scheme(house,seventeenth,eighteenth,nineteenth,twentieth), listing the number of monarchs of each royal house that acceded to the throne in the 17th,18th,19th and 20th centuries and have an issue of what to add as my THEN query.
EDIT:
Thank you for your suggestions! I made some changes to my query now:
SELECT house,
TO_CHAR(accession, 'YYYY' ) AS accession_year,
COUNT(CASE WHEN accession_year BETWEEN 1601 AND 1700 THEN name END) AS seventeenth,
COUNT(CASE WHEN accession_year BETWEEN 1701 AND 1800 THEN name END) AS eighteenth,
COUNT(CASE WHEN accession_year BETWEEN 1801 AND 1900 THEN name END) AS nineteenth,
COUNT(CASE WHEN accession_year BETWEEN 1901 AND 2000 THEN name END) AS twentieth,
FROM monarch
WHERE house IS NOT NULL
GROUP BY house
;
Now psql tells me that accession_year does not exist. I do not want to use full accession date in CASE statements. How can I still use my TO_CHAR in the query?

Answer 1

Each expression in a single SELECT clause is evaluated "as if" it's being computed in parallel with all other expressions in the same clause. As such, you're not allowed to have any dependencies between them since no resulting values are available at the start.
One option is to introduce a subquery:
SELECT house,
accession_year,
COUNT(CASE WHEN accession_year BETWEEN 1601 AND 1700 THEN name END) AS seventeenth,
COUNT(CASE WHEN accession_year BETWEEN 1701 AND 1800 THEN name END) AS eighteenth,
COUNT(CASE WHEN accession_year BETWEEN 1801 AND 1900 THEN name END) AS nineteenth,
COUNT(CASE WHEN accession_year BETWEEN 1901 AND 2000 THEN name END) AS twentieth,
FROM (
SELECT house,name,TO_CHAR(accession, 'YYYY' ) AS accession_year
FROM monarch
WHERE house IS NOT NULL ) AS t
GROUP BY house
;
Where now you have two separate SELECT clauses and the outer one is allowed to depend on values computed by the inner one.

Answer 2

You can use sub-selects OR CTEs to SELECT (or just use) calculated columns, but in some simpler cases (like yours) a LATERAL join is more readable:
SELECT house,
COUNT(CASE WHEN accession_century = 17 THEN 1 END) AS seventeenth,
COUNT(CASE WHEN accession_century = 18 THEN 1 END) AS eighteenth,
COUNT(CASE WHEN accession_century = 19 THEN 1 END) AS nineteenth,
COUNT(CASE WHEN accession_century = 20 THEN 1 END) AS twentieth
FROM monarch,
date_part('century', accession) accession_century
WHERE house IS NOT NULL
GROUP BY house
Note: to_char() returns a string, which is not really useful for querying, use date_part() or EXTRACT() instead. Especially in your case: they have the ability to extract the century, which you want to search for.