Alias Reference Date_Diff Days. Need to Parse or create temp table with dates? - sql

Question

Below I have the tables and query which output the below
Table1
EmployeeID | StartDateTimestamp | CohortID | CohortName
---------- | ------------------ | -------- | ----------
1 | 20080101 01:30:00 | 1 | Peanut
1 | 20090204 01:01:00 | 2 | Apple
2 | 20190107 05:52:14 | 1 | Peanut
3 | 20190311 02:35:26 | 2 | Apple
Employee
EmployeeID | HireStartName | StartDateTimestamp2
---------- | ------------- | -------------------
1 | HiredStart | 20080501 01:30:00
1 | DeferredStart | 20090604 01:01:00
2 | HiredStart | 20190115 05:52:14
3 | HiredStart | 20190330 02:35:26
Query
select
t.cohortid,
min(e.startdatetimestamp2) first,
max(e.startdatetimestamp2) last
from table1 t
inner join employee e on e.employeeid = t.employeeid
group by t.cohort_id
Output
ID | FIRST | LAST
1 |20190106 12:00:05 |20180214 03:45:12
2 |20180230 01:45:23 |20180315 01:45:23
My attempt:
SELECT DATE_DIFF(first, last, Day), ID, max(datecolumn1) first, min(datecolumn1) last
Error: Unrecognized name.
How do I enter the reference alias first and last in a Date_Diff?
Do I need to derive a table?
Clarity: Trying to avoid inputting in the dates, since I am looking to find the date diff of both first and last columns for as many rows as there is data.
This answer has been discussed here: Date Difference between consecutive rows
DateDiff has deprecated, and now it is Date_Diff (first, last, day)
Then I tried:
SELECT ID, DATE_DIFF(PARSE_DATE('%y%m%d',t.first), PARSE_DATE('%y%m%d',t.last), DAY) days
FROM table
Failed to parse input string "20180125 01:00:05"
Tried this
SELECT CohortID, date_diff(first,last,day) as days
FROM (select cohortid,min(startdatetimestamp2) first,
max(startdatetimestamp2) last
FROM employee
JOIN table1 on table1.employeeid = employee.employeeid
group by cohortid)
I get days not found on either side of join

Answer 1

Regarding your first question about using aliases in a query, there are some restriction where to use them, specially in the FROM, GROUP BY and ORDER BY statements. I encourage you to have a look here to check these restrictions.
About your main issue, obtaining the date difference between two dates. I would like to point that your timestamp data, in both of your tables, are actually considered as DATETIME format in BigQuery. Therefore, you should use DATETIME builtin functions to get the desired results.
The below query uses the data you provided to obtain the aimed output.
WITH
data AS
(
SELECT
t.cohortid AS ID,
PARSE_DATETIME('%Y%m%d %H:%M:%S', MIN(e.startdatetimestamp2)) AS first,
PARSE_DATETIME('%Y%m%d %H:%M:%S', MAX(e.startdatetimestamp2)) AS last
FROM
`test-proj-261014.sample.table1` t
INNER JOIN
`test-proj-261014.sample.employee` e
ON
e.employeeid = t.employeeid
GROUP BY t.cohortid
)
SELECT
ID,
first,
last,
DATETIME_DIFF(last, first, DAY ) AS diff_days
FROM
data
And the output:
Notice that I created a temp table to format the fields StartDateTimestamp and StartDateTimestamp2, using the PARSE_DATETIME(). Afterwards, I used the DATETIME_DIFF() method to obtain the difference in days between the two fields.