SQL 计数匹配

Question

I have a list of 191 values that I want to compare to a column. I ultimately want to get a count of % of rows that have a value in my master list ( matches/(non-matches + NULL) ).

I know I could do something like below but am wondering if this is the most efficient way? Is it possible to create an array that stores the values and check against this? Not sure what the best practice way of doing this is given I have 191 values to check against?

I am hoping to avoid dropping 191 csv into the argument since this is a hit to formatting/readability. Is there a way to store these values inside an array or temp table so i can just drop a short hand variable into the actual query? Or is using the below method/average method still the best way to go regardless of how many values there are to check against?

SELECT
    SUM(CASE WHEN COALESCE(field, '') IN (COMMA SEPARATED VALUES) THEN 1 ELSE 0 END) as matches,
    COUNT(COALESCE(field)) as total_rows
FROM table

---

Also, I believe that COUNT(*) and COUNT(1) are blind to NULL fields so can anyone confirm if using COUNT(COALESCE(FIELD)) ensures the count includes null values from the field?

Answer 1

Presto doesn't support temporary tables, but you could improve readability of your query by using an inline table ( WITH clause combined with VALUES ) to avoid having a long list of values in the aggregation expression.

Then, you can count the number of matches by doing the following. Note the use of FILTER for improved readability.

count(*) FILTER (WHERE field IN (SELECT value FROM data))

Here's a complete example:

WITH data(value) as (VALUES
   'value1',
   'value2',
   ...
)
SELECT
    count(*) AS total_rows,
    count(*) FILTER (WHERE field IN (SELECT value FROM data)) AS matches
FROM t

Answer 2

I think you just want:

SELECT AVG(CASE WHEN field IN (COMMA SEPARATED VALUES) THEN 1.0 ELSE 0 END) as match_ratio
FROM table

Answer 3

One option uses avg() :

select avg(case when field in (<<csv list>>) then 1.0 else 0 end) rows_ratio
from mytable

It might be simpler to use an array to pass the values:

select avg(case when contains(<< array of values>>, field) then 1.0 else 0 end) rows_ratio
from mytable