cancel
Showing results for
Did you mean:

## Aggregating Duration/Time

This is a collaborative blog post by @konstantinos and @Greg_Deckler resulting from the forum discussion topic “Aggregating Duration/Time”. http://community.powerbi.com/t5/Desktop/Aggregating-Duration-Time/m-p/13350/highlight/true#M3358

Introduction

Both Excel and DAX lack support for a true time duration data type making the display of duration in a format such as “HH:MM:SS” problematic. However, with some creative formulas, this problem can be solved.

Scenario

Consider a scenario where you have a number of seconds worked on a help desk ticket or the duration of a phone call in number of seconds. You desire to put this into a more standardized time duration format such as “HH:MM:SS”.

Here is a small sample of data to work with:

Item,Duration in Seconds

Phone Call 1, 45

Phone Call 2, 2875

Phone Call 3, 8944

Excel

In Excel, we would have the Item descriptions in column 1, A and Duration in Seconds in column 2, B. We could create the following additional columns to solve the problem:

• Hours (C column): = INT(B2/3600)
• Minutes (D column): = INT(MOD(B2,3600)/60)
• Seconds (E column): = MOD(MOD(B2,3600)/60)

However, these calculations do not account for leading zeros and we might end up with a duration of “4:2:8” for four hours, two minutes and eight seconds. Not exactly what we want. So, we have to add more columns:

• H (F column): =IF(LEN(C2)=1,CONCATENATE("0",C2),CONCATENATE("",C2))
• M (G column): =IF(LEN(D2)=1,CONCATENATE("0",D2),CONCATENATE("",D2))
• S (H column): =IF(LEN(E2)=1,CONCATENATE("0",E2),CONCATENATE("",E2))

For Duration, we cannot use the TIME function of Excel as this would return a point in time, such as 12:47 AM. Therefore, we must use CONCATENATE:

• Duration (I column): =CONCATENATE(F2,":",G2,":",H2)

Now we have the Duration in the desired format. Unfortunately, we also have six extra columns cluttering things up. We could hide those columns in the Excel interface or we might even combine everything into a single formula:

• Duration: =CONCATENATE(IF(LEN(INT(B2/3600))=1,CONCATENATE("0",INT(B2/3600)),CONCATENATE("",INT(B2/3600))),":",IF(LEN(INT(MOD(B2,3600)/60))=1,CONCATENATE("0",INT(MOD(B2,3600)/60)),CONCATENATE("",INT(MOD(B2,3600)/60))),":",IF(LEN(MOD(MOD(B2,3600),60))=1,CONCATENATE("0",MOD(MOD(B2,3600),60)),CONCATENATE("",MOD(MOD(B2,3600),60))))

If you are a fan of this second option, you have likely spent far too much time writing Perl code and need to seek professional counseling.

The problem is that both of these options are sub-optimal. When someone else looks at what has been done, or even if you were to look at this 6 months from now, you would essentially have to reverse engineer what has been done, which could take a significant amount of time and troubleshooting. For example, what if one of the “3600” had been mistakenly entered as “360” and was causing a problem? Tracking that down could take some time.

DAX

There are so many similarities between DAX and Excel that our functions are nearly identical. Given the same data and the columns [Item] and [Duration in Seconds], our initial formulas are identical other than the naming convention for the cell/column:

• Hours = INT([Duration in Seconds]/3600)
• Minutes = INT(MOD([Duration in Seconds],3600)/60)
• Seconds = MOD(MOD([Duration in Seconds],3600),60)
• H = IF(LEN([Hours])=1,CONCATENATE("0",[Hours]),CONCATENATE("",[Hours]))
• M = IF(LEN([Minutes])=1,CONCATENATE("0",[Minutes]),CONCATENATE("",[Minutes]))
• S = IF(LEN([Seconds])=1,CONCATENATE("0",[Seconds]),CONCATENATE("",[Seconds]))

One difference between DAX and Excel however is that DAX’s CONCATENATE function only accepts two arguments, so the final Duration column formula becomes:

• Duration = CONCATENATE([H],CONCATENATE(":",CONCATENATE([M],CONCATENATE(":",[S]))))

Just like Excel, we could combine all of those formulas into a single massive formula, but we have no desire to make your eyes bleed. Thus, we are left in the same predicament as Excel when it comes to maintainability, supportability and overall readability of the solution. Or are we?

A Better Solution in DAX

Luckily, the powerful language-like features of DAX, comments and variables, help us improve our solution dramatically.

Consider the following DAX “formula”:

```Duration =
// Duration formatting
// * @konstatinos 1/25/2016
// * Given a number of seconds, returns a format of "hh:mm:ss"
//
VAR Duration = [Duration in Seconds]
// There are 3,600 seconds in an hour
VAR Hours =
INT ( Duration / 3600)
// There are 60 seconds in a minute
VAR Minutes =
INT ( MOD( Duration - ( Hours * 3600 ),3600 ) / 60)
// Remaining seconds are the remainder of the seconds divided by 60 after subtracting out the hours
VAR Seconds =
ROUNDUP(MOD ( MOD( Duration - ( Hours * 3600 ),3600 ), 60 ),0) // We round up here to get a whole number
// These intermediate variables ensure that we have leading zero's concatenated onto single digits
VAR H =
IF ( LEN ( Hours ) = 1,
CONCATENATE ( "0", Hours ),
CONCATENATE ( "", Hours )
)
VAR M =
IF (
LEN ( Minutes ) = 1,
CONCATENATE ( "0", Minutes ),
CONCATENATE ( "", Minutes )
)
VAR S =
IF (
LEN ( Seconds ) = 1,
CONCATENATE ( "0", Seconds ),
CONCATENATE ( "", Seconds )
)
// Now return hours, minutes and seconds with leading zeros in the proper format "hh:mm:ss"
RETURN
CONCATENATE (
H,
CONCATENATE ( ":", CONCATENATE ( M, CONCATENATE ( ":", S ) ) )
)```

Conclusion

The techniques shown here allow us to solve the problem of the lack of a true time duration data type in both Excel and DAX. Ultimately, DAX allows us to create a superior solution due to its support for powerful programming-language-like features such as comments and variables without resorting to a completely different programming language such as VBScript.

@Greg_Deckler and @konstantinos keep the good stuff coming! Thanks guys!

Instead of doing that extensive concatenation, how about that short classic ...

H = RIGHT(100+[Hours],2) M = RIGHT(100+[Minutes],2) S = RIGHT(100+[Seconds],2)

Really Helpful. Especialy working with large times!!

Thanks

Anonymous

Thanks for the great post!

I have a two-column table: "Date", and "Average Length of Call in Seconds".  I've created a new column using your code above which works perfectly.  But I end up getting total number of seconds for the Y-Axis labels.

Is there any way to show duration for the Y-Axis labels in mm:ss format?  If so, can you point me in the right direction?

Thanks for any help!

--Michael--

Hello,

Thanks for this formula. I've gone through a similar process using Tableau (which also lacks a handy 'time duration' feature). However, I'm not able to run calculations on this since the end result is a concatenated string. How would I display the average Duration? Or, supposing I wanted to show a trend over time. How would I display Duration MoM?

Thanks!

So, today you would have to perform the calculations while it is in a number format like total seconds and then convert it at the end. I realize that this does not help with thing like column charts and such. I just returned from the Microsoft MVP Summit in Bellevue and I specifically brought this issue to the attention of the Power BI Desktop team, Gateways and Connectors team and the SSAS Tabular team and believe I found an owner for the issue and so hopefully (fingers crossed) the issue will get sorted out soon. The problem is this issue lives across all 3 teams, the duration data type needs to be supported in Power Query (which it is, Time.Duration) but then also in the tabular data model (not currently supported) and then it would need a supported user interface (formatting) in the Desktop (can't do this until the data model supports it. I had some good discussions with the SSAS Tabular team about how to store it in the data model (as seconds, milliseconds, etc.) So, again, hopefully we will see something soon. Definitely vote for the Idea!

Hi there

how to do this with Minutes?

For example:
I have 2.480,08 minutes

I want to format this in duration (but not in Time-Format) because the value is > than 24 hours.

Result shoud be:

1 day 08:22:16

But without VAR - this slowdown extremly all my visualizations - isn´t it?

How to proceed?

HI Smoupre,

This helped, although trying to find a easier solution I figured below might help

First convert the data into seconds.  Then divide the data by 86400 (60 min x 60 secs x 24 hours) to get output that can be converted to time.  Then use below to get time equivalent

FORMAT([seconds]/86400,"Long Time").  This gives time equivalent with AM / PM at end.  You can use Left to trim it.

Left(FORMAT([seconds]/86400,"Long Time"),7)

This solves the averaging problem and other time duration related problems.

Additional Date/Time formats in DAX can be find below

https://technet.microsoft.com/en-us/library/ee634813(v=sql.105).aspx

Hope this helps.

Thanks

Disclaimer - I am new to DAX and PowerBi so my skills are rudemantary.

This is taking forever unfortunately 😞

When used in a matrix it gives the first or the last instances, what is the work around for that?

New here and thank you for posting that formula for conversion! One question....I'm converting milliseconds so would I only need to change the first part of the formula for VAR Hours? From 3600 to 3600000. Or, would I have to change everywhere I see 3600 to 3600000?

Hi there 🙂 Thanks for this @Greg_Deckler ! Very helpful since I work from home so I can't use the Power Query Editor as I don't have access to our MySQL DB.

But I was wondering, do you know how I could change the data type so that I could plot graphs showing for example, Time To Repar by Machine?

At the moment the data is in text format so I can't do this, and when I try formatting it as time, it expects the highest "hour" value to be 23...

@LiziM Can't be completely certain of your situation, but see Chelsie Eiden's Duration:

https://community.powerbi.com/t5/Quick-Measures-Gallery/Chelsie-Eiden-s-Duration/m-p/793639#M389

Y'all are making this overly complicated.

The FORMAT function can do most of the work here:

``````Duration =
VAR _Seconds = [Sec]
VAR _Hours = INT ( _Seconds / 3600 )
VAR _Remainder = ( _Seconds - _Hours * 3600 ) / 3600 / 24
RETURN
FORMAT ( _Hours, "00:" ) & FORMAT ( _Remainder, "nn:ss" )``````

Note: FORMAT assumes durations are in units of days, so a seconds --> days conversion is included in the _Remainder variable.

Top Kudoed Posts
Latest Articles
Archives