Running median last n days group by user. BigQuery SQL - sql

Question

I have a table like this:
date
user_id
revenue
2021-10-01
1
2
2021-10-02
1
3
2021-10-05
1
2
2021-10-09
1
3
2021-10-15
1
3
2021-10-01
2
2
2021-10-04
2
2
2021-10-10
2
1
2021-10-11
2
3
2021-10-11
2
3
2021-10-20
2
5
And I want to add column with median revenue for the last 5 days group by user_id. Desired output should looks like this:
date
user_id
revenue
median_last_5_days
2021-10-01
1
2
NULL
2021-10-02
1
3
2
2021-10-05
1
2
2.5
2021-10-09
1
3
2
2021-10-16
1
3
NULL
2021-10-01
2
2
NULL
2021-10-02
2
3
2
2021-10-03
2
3
2.5
2021-10-04
2
2
3
2021-10-10
2
1
NULL
2021-10-11
2
3
1
2021-10-11
2
3
2
2021-10-20
2
5
NULL
Can I produce this with SQL BigQuery?
Thanks for helping me)

Answer 1

Use of PERCENTILE_CONT or PERCENTILE_DISC to get the median will not work on these conditions as window_frame_clause are not allowed in Navigation functions.
Try this approach,
CREATE TEMP FUNCTION MEDIAN(arr ANY TYPE) AS ((
SELECT
IF(
MOD(ARRAY_LENGTH(arr), 2) = 0,
(arr[OFFSET(DIV(ARRAY_LENGTH(arr), 2) - 1)] + arr[OFFSET(DIV(ARRAY_LENGTH(arr), 2))]) / 2,
arr[OFFSET(DIV(ARRAY_LENGTH(arr), 2))]
)
FROM (SELECT ARRAY_AGG(x ORDER BY x) AS arr FROM UNNEST(arr) AS x)
));
SELECT
date,
user_id,
revenue,
MEDIAN(
ARRAY_AGG(revenue) OVER (PARTITION BY user_id
ORDER BY datetime_diff(date, date('2000-01-01'), day)
RANGE BETWEEN 4 PRECEDING AND 1 PRECEDING)
) AS median_last_5_days
FROM `my-project.my-dataset.my-table`
GROUP BY date, user_id, revenue
ORDER BY user_id;
Sample data:
Output: