I'm trying to create a custom query in Tableau to use on Google's BigQuery. The goal is to have an offset parameter in Tableau that changes the offsets used in a date based WHERE clause.
In Tableau it would look like this:
SELECT
DATE_ADD(UTC_USEC_TO_MONTH(CURRENT_DATE()),<Parameters.Offset>-1,"MONTH") as month_index,
COUNT(DISTINCT user_id, 1000000) as distinct_count
FROM
[Orders]
WHERE
order_date >= DATE_ADD(UTC_USEC_TO_MONTH(CURRENT_DATE()),<Parameters.Offset>-12,"MONTH")
AND
order_date < DATE_ADD(UTC_USEC_TO_MONTH(CURRENT_DATE()),<Parameters.Offset>-1,"MONTH")
However, BigQuery always returns an error:
Error: DATE_ADD 2nd argument must have INT32 type.
When I try the same query in the BigQuery editor using simple arithmetic it fails with the same error.
SELECT
DATE_ADD(UTC_USEC_TO_MONTH(CURRENT_DATE()),5-3,"MONTH") as month_index,
FROM [Orders]
Any workaround for this? My only option so far is to make multiple offsets in Tableau, it seems.
Thanks for the help!
I acknowledge that this is a hole in functionality of DATE_ADD. It can be fixed, but it will take some time until fix is rolled into production.
Here is a possible workaround. It seems to work if the first argument to DATE_ADD is a string. Then you can truncate the result to a month boundary and convert it from a timestamp to a string.
SELECT
FORMAT_UTC_USEC(UTC_USEC_TO_MONTH(DATE_ADD(CURRENT_DATE(),5-3,"MONTH"))) as month_index;
Related
The query I'm running as a test is:
SELECT
UNIX_DATE(created_utc)
FROM `fh-bigquery.reddit_comments.2017_08`
But I keep getting this error:
Error: No matching signature for function UNIX_DATE for argument types:
INT64. Supported signature: UNIX_DATE(DATE) at [2:3]
I checked the datatype for the created_utc field and it's an integer. Casting and whatnot won't work either.
Would really appreciate any help. Thanks!
You should use TIMESTAMP_SECONDS() instead
#standardSQL
SELECT
TIMESTAMP_SECONDS(created_utc)
FROM `fh-bigquery.reddit_comments.2017_08`
LIMIT 5
Then you can use DATE() if you need date only
DATE(TIMESTAMP_SECONDS(created_utc))
UNIX_DATE() takes a String.
And DATE_FROM_UNIX_DATE() takes an INT64. SQL has a legacy problem of thinking of time ("date") in DAYS and not SECONDS like Unix. Thus:
SELECT DATE_FROM_UNIX_DATE(CAST(created_utc/86400 as INT64))
FROM `fh-bigquery.reddit_comments.2017_08`
I have a SQL query which shows time activity of each account. Database is Microsoft SQL Server on Windows Server 2008.
Help me please to translate this query to tableau with using parameters Parameters.Date1 and Parameters.Date2 instead of #time.
The result of the query:
USER,Date,Total time
USER1,2016-09-22,07:00:00.0000000
USER2,2016-09-22,08:00:00.0000000
USER3,2016-09-22,05:00:00.0000000
SQL query:
DECLARE #time datetime
set #time = '08.09.2016'
SELECT
[User],
CAST(DATEADD(SECOND, sum(datediff(DAY, #time, [Start])), #time) AS date) 'Date',
CAST(DATEADD(SECOND, sum(datediff(SECOND, '00:00:00',[Period])), '00:00:00') AS time) 'Total time'
FROM
[User].[dbo].[UserAction]
WHERE
[Start] >= #time+'00:00:00' and [Start] <= #time+'23:59:59'
GROUP BY
[USER]
input data to build the query:
USER, Start,End,Period
USER1,2016-09-22 09:00:00.000,2016-09-22 12:00:00.000,03:00:00
USER1,2016-09-22 12:00:00.000,2016-09-22 13:00:00.000,01:00:00
USER1,2016-09-22,13:00:00.000,2016-09-22 16:00:00.000,03:00:00
USER2,2016-09-22,09:00:00.000,2016-09-22 13:00:00.000,04:00:00
USER2,2016-09-22,13:00:00.000,2016-09-22 17:00:00.000,04:00:00
USER3,2016-09-22,09:00:00.000,2016-09-22 10:00:00.000,01:00:00
USER3,2016-09-22,10:00:00.000,2016-09-22 12:00:00.000,02:00:00
USER3,2016-09-22,12:00:00.000,2016-09-22 14:00:00.000,02:00:00
I don't have enough imaginary stack overflow points yet to make a comment instead of an answer, but I would agree with Gordon Linoff.
A table valued function in sql can be used directly in a Tableau data source, and it's treated just like a table.
Note I did not test the below, but here is what the equivalent function might look like:
CREATE FUNCTION dbo.MyFuntion (#time datetime)
RETURNS TABLE
AS
RETURN
(
SELECT
[User]
,cast(DATEADD(SECOND, sum(datediff(DAY, #time,[Start])),#time) as date)'Date'
,cast(DATEADD(SECOND, sum(datediff(SECOND, '00:00:00',[Period])),'00:00:00') as time)'Total time'
FROM
[User].[dbo].[UserAction]
WHERE
[Start] >= #time+'00:00:00' and [Start] <= #time+'23:59:59'
GROUP BY [USER]
);
Tableau 9 (haven't tried 10) seems to discourage custom SQL (it warns anyone that opens your workbook) and stored procedures (slow vs. same sql in a function).
Alternatively, adding the pure dbo.UserAction table to a data source and making calculated fields for the second two columns might work: Tableau Documentation. It seems to have all the functions needed to manipulate dates. However, there may be some crazy limitation associated with parameters that might limit it, honestly can't remember off the top of my head.
You don't need custom SQL for this. Keep it simple. Connect Tableau directly to your UserAction table.
You can either:
Put Day(Start) on the filter shelf, Make sure it is a continuous Date truncated to the Day. Show the filter and set the filter to let you pick a single value at a time - I would choose a slider UI.
Or write a calculated field to put on the filters shelf that references a parameter such as day(Start) = day(Date1)
Put User on one shelf, such as rows, and Sum(Period) on another such as columns. That should do it unless Tableau has trouble interpreting your Period field datatype. If so, try changing the datatype to Number inside Tableau to see if it converts durations to numbers automatically, if not you may need to write a calculated field for the conversion.
I'm working in an aviation environment where I have to compare two values. The OEM in all its wisdom has stored all datetime values in char. The values to compare are called, say, aibt and sibt
As part of the where clause I need to check if aibt is greater than sibt. So this is what I did :
To_Date(aibt,'YYYYMMDDHH24MISS') > To_Date(sibt,'YYYYMMDDHH24MISS')
When applying to_date, the original char values are converted to : 25.12.2013 12:54:00
But I get the following error :
ORA-01841: (full) year must be between -4713 and +9999, and not be 0
You obviously have some data that can not be converted with TO_DATE to a proper date. Feel free to berate your OEM for not defining constraints or otherwise prevent this from happening.
While he gets around to fix his problems, you can use this as a workaround:
WHERE aibt > sibt
Seems like an error in your data. Try this first:
Select to_date(aibt,'YYYYMMDDHH24MISS') from your_table
Select to_date(sibt,'YYYYMMDDHH24MISS') from your_table
to check if you have some incorrect data in your columns.
When you identified your problematic data in your column, you could find the data that is messing this up with somethin like this:
select *
from yourtable
where not regexp_like(aibdt, '^[[:digit:]]+$')
OR length(aibdt)<4 OR
substr(aibdt,1,4)<0
There was some problem in the data. Also thanks Twinkles for your contribution. I hadnt thought of that. Can anyone explain how comparing two char values work?
I got the answer (what I'd really required) in another post Oracle: how to add minutes to a timestamp? which I used for constructing the comparison :
WHERE To_Date(aibt,'YYYYMMDDHH24MISS')>To_Date(sibt,'YYYYMMDDHH24MISS')+ INTERVAL '15' MINUTE
Thanks everyone!
I have a query in which I want to select data from a column where the data is a date. The problem is that the data is a mix of text and dates.
This bit of SQL only returns the longest text field:
SELECT MAX(field_value)
Where the date does occur, it is always in the format xx/xx/xxxx
I'm trying to select the most recent date.
I'm using MS SQL.
Can anyone help?
Try this using ISDATE and CONVERT:
SELECT MAX(CONVERT(DateTime, MaybeDate))
FROM (
SELECT MaybeDate
FROM MyTable
WHERE ISDATE(MaybeDate) = 1) T
You could also use MAX(CAST(MaybeDate AS DateTime)). I got in the (maybe bad?) habit of using CONVERT years ago and have stuck with it.
To do this without a conversion error:
select max(case when isdate(col) = 1 then cast(col as date) end) -- or use convert()
from . . .
The SQL statement does not specify the order of operations. So, even including a where clause in a subquery will not guarantee that only dates get converted. In fact, the SQL Server optimizer is "smart" enough to do the conversion when the data is brought in and then do the filtering afterwards.
The only operation that guarantees sequencing of operations is the case statement, and there are even exceptions to that.
Another solution would be using PATINDEX in WHERE clause.
SELECT PATINDEX('[0-9][0-9]/[0-9][0-9]/[0-9][0-9][0-9][0-9]', field_value)
Problem with this approach is you really are not sure if something is date (e.g. 99/99/9999 is not date).
And problem with IS_DATE is it depends on configuration (e.g. DATEFORMAT).
So, use an appropriate option.
i using MySQL Query for my task.
And I interested using Date and time function.
can i use DAY(), WEEK(), and YEAR() at one query?
SELECT Object
FROM table
WHERE DAY(date) BETWEEN 1 AND 7
GROUP BY WEEK(date, 1), YEAR(date)
i want do this bcoz i'm worry if sometimes my program have an error because of the date setting and not recognize some date.please give me an input.
Yes, you can use them all in a single query.
The only disadvantage I can think of is that using any of the DAY, WEEK or YEAR functions won't be able to use the index on the column the function is applied to, assuming one is present.
If you're having issues relating to date formatting, you should get familiar with:
DATE_FORMAT
STR_TO_DATE