Hi Taron,
I don’t see any dependency on day of week with my approach. Let me know if I’m missing something.

Hi Steven/Phil,
in this approach , what if its last sunday of month . than what will be n value ? and any new way of sorting this issue out

Stanley, I’m not sure what you’re asking. What can we clarify? Thanks!

How do I use your function?

I just found DateTimeZone.ToLocal() is NOT working on Power Bi Service.

Desktop is in AEST (Australian Eastern Standard Time) which is my local time.

However, as soon as I refresh the PBI service all dates revert back to UTC :(

Abram, many thanks for your comment!

I’ve tested myself and confirm that DateTimeZone.ToLocal() is now using the client time zone and not that of the servers running the service

I’ll update the post as well.

Hi Hendrik,

With my approach, you create a local time column in Power Query and execute a function against the column.

With Steven’s approach, you are using a calculated column.

There are pros and cons with each. My approach calculates the values at data load/refresh time. Steven’s approach calculates as needed/when used.

Regards,
Phil

How do I implement Steven’s solution?

I have a table with one column being UTC. I need a new column with PST / PDT.

Do I put his code in a calculated column? Or create a function that gets called by the column?

Thanks for the post Phil & the subsequent post Steven. Just what i needed.

Interesting approach. Thanks!

Last Refresh Pacific =

// First let’s find the date of the second Sunday of March
// For the formula below:
// In DAX, we use call DATE() and WEEKDAY()
// “1+7*2”, generically, “1+7*n” where n represents
// the nth occurrence of that “weekday name”
// “8-1”, generically, “8-y”, where
// y is a number from 1 to 7, where
// 1 = Sunday through to 7 = Saturday
//
// Thus, the date of the second Sunday of the month
// is returned to the variable ZZQQ_DST_START
//
// ZZQQ_DATE is the date/time in UTC.
//
VAR ZZQQ_DST_START = DATE(YEAR(ZZQQ_DATE),3,1+7*2)
– WEEKDAY(DATE(YEAR(ZZQQ_DATE),3,8-1))

// Next let’s find the date of the first Sunday in November
VAR ZZQQ_DST_END = DATE(YEAR(ZZQQ_DATE),11,1+7*1)
– WEEKDAY(DATE(YEAR(ZZQQ_DATE),11,8-1))

RETURN

IF (
// Evaluate if ZZQQ_DATE is between DST start and DST end
// If yes, change time to Pacific Daylight Savings Time
// else, change time to Pacific Standard Time
(ZZQQ_DATE >= ZZQQ_DST_START) && (ZZQQ_DATE <= ZZQQ_DST_END),
ZZQQ_DATE + (-7/24), // If condition matches, convert UTC to PDT
ZZQQ_DATE + (-8/24) // Else convert UTC to PST
)

Hi Kaden,
Glad you found the post helpful. Regarding the tables, consider implementing a single Dates table that is referenced via a date key, and only add the columns to the Dates table.

Great and insightful post. However, I am wondering if you have been able to reach a solution now in 2018 where you do not have to add a custom column. Our team has several date columns and to create a duplicate of each column to invoke the custom function will add a lot of heaviness to our tables since they are already very large. Have you been able to find a leaner solution since your post? Thanks!