SQL: CASE WHEN having AVG() as condition not giving right output - sql

Question

I have a table of unique users that each has a "rating" column (it's an average rating they give out of all their ratings given in a different table of reviews). I want to add another column to my table, which specifies either them giving a rating that is above the average of all ratings of all users (hence I use the AVG() function), below or at average (I call it "bias"). In other words, I want to see whether each user gives on average higher or lower ratings than the total average. I understand the limitedness of this query, and ideally I would include an interval (i.e. within 0.5 points below or above average still counts as average) but I can't seem to make even the simplest query work.
I've been using the Yelp dataset from a Coursera course, but I tried to create a sample that produces the same result that I do not want - just one row. I want to have this categorization for each row, hence it should return 3 rows in this example, "below average" in the first two and "above average" in the third. However, the code below produces just one row. I have been working with R and this seems like I am using incorrect syntax, but after 30 minutes of searching the web I cannot find a solution.
I am working in and want to use SQLite syntax as part of the course in Coursera
CREATE TABLE test
(
id integer primary key,
rating integer
);
INSERT INTO test
(id, rating)
VALUES
(1, 1);
INSERT INTO test
(id, rating)
VALUES
(2, 3);
INSERT INTO test
(id, rating)
VALUES
(3, 8);
SELECT id,
rating,
CASE
WHEN rating > AVG(rating) THEN "above average"
WHEN rating < AVG(rating) THEN "below average"
ELSE "no bias"
END AS "bias"
FROM test

Answer 1

You can't use the aggregate function AVG() like this.
But you can do it with AVG() window function:
SELECT id,
rating,
CASE
WHEN rating > AVG(rating) OVER () THEN "above average"
WHEN rating < AVG(rating) OVER () THEN "below average"
ELSE "no bias"
END AS "bias"
FROM test
See the demo.
Results:
| id | rating | bias |
| --- | ------ | ------------- |
| 1 | 1 | below average |
| 2 | 3 | below average |
| 3 | 8 | above average |

Answer 2

SELECT id,
rating,
CASE
WHEN rating > (select AVG(rating) from test) THEN "above average"
WHEN rating < (select AVG(rating) from test) THEN "below average"
ELSE "no bias"
END AS "bias"
FROM test
AVG is an aggregate function and works in conjunction with a GROUP BY.
when you do not specify anything in the GROUP BY section, it will aggregate the whole table thus reducing it to one row.
Generally you select aggregated columns and non-aggregated columns without specifying the non aggregated columns in the GROUP BY list. I am not a big fan of DBMS which allow this behavior (SQLLite seems to be an offender).
What I did in the query above is that I calculated the average of the whole table using a subquery. And then compared each row against the average.
Or like others have specified you can go with WINDOW functions. Where you apply a function over a some parts of the data as defined by your window. They look like theri regular aggregated functions conunterparts but you will notice the OVER keyword which specify they are applied over a window. In the over clause you can partition your data or you can use it as a whole. For example if you had multiple stores and sales amount per day for each store, you could parition by store to compute the per store average.