0:00Hey it's Tim here, in today's video we're
0:01talking about the attribute function inside
0:04of Tableau.
0:04In one of my previous videos where I was
0:06talking about new models in Tableau 2020.4,
0:09I used the attribute function to get me out
0:11of a tight spot inside of a calculation.
0:13And so a subscriber simply asked, "hey can
0:15you go into more detail about that function
0:17?" So here we
0:18are. In today's video what I'm going to do
0:20is tell you what the attribute function is,
0:22some use cases
0:24for it, and then lastly some limitations to
0:26be aware of. Okay let's get stuck in. Okay
0:28to get this
0:28video started it's really important you pay
0:30attention to this next step. I need you to
0:32connect
0:32to the American Superstore sales data set.
0:35That's the second one here on the bottom
0:37left. Okay
0:38as long as you connect into the American
0:39one this is going to work because
0:41essentially I need the
0:42fact that there are cities in the United
0:44States that the name exists in multiple
0:46states. You'll
0:47see what I mean by that in a little bit.
0:50Okay so when we're working with Tableau and
0:53you want to
0:53know what an attribute function is or you
0:55want to know what any function is, you can
0:57actually just
0:58look inside of the product itself to find
0:59out the meaning. I'm going to go ahead and
1:01open the
1:02calculation window here, create a
1:03calculated field and you'll notice that
1:06mine opens up with this
1:07side pane enabled already. Now if you can't
1:09see this don't worry there's this tiny
1:11little arrow
1:11here. This is always frustrating for new
1:13users because when they start out with
1:15Tableau it opens
1:16this window and you don't necessarily know
1:18that there's more to offer when you click
1:20this button.
1:20But when you click this button essentially
1:22you get this list and these are all the
1:24functions that are
1:24available in Tableau. Now at the very top
1:26they're actually grouped okay so you can
1:29see all the
1:29different groupings for these functions. If
1:32I go down to this aggregate grouping you
1:34can see that
1:35the attribute one is right at the top so
1:37that's the first clue. The attribute
1:39function is a type
1:40of aggregation inside of Tableau. What does
1:42it do? Well there's a description right
1:44here so returns
1:45a value of the given expression if it only
1:48has a single value for all the rows in the
1:50group.
1:51Otherwise it displays an asterisk. Null
1:53values are ignored okay so that's
1:55essentially what it does.
1:56Let's take a look at that and I'm just
1:58going to put this here to the bottom so we
2:00can sort of keep
2:00it in view whilst we work with it. We're
2:03also going to work in the data viz pane
2:04whilst that's open. So
2:06I'm going to go ahead and go to location. I
2:08'm going to open up the hierarchy by
2:09clicking this
2:10down arrow here. Then I'm going to drag
2:13city onto rows then state onto rows as well
2:16and what we'll
2:17get is this table. It's a very simple table
2:20that has city on the left state on the
2:22right. Now you
2:23will notice that there are some cities here
2:25that have multiple states. This makes sense
2:28because
2:28Arlington is both the name of a city in
2:30Texas and in Virginia and so on and so
2:33forth. So you get the
2:34idea and Obern exists in three states as a
2:37name. So now we've got this set up
2:40correctly we can
2:41actually start to see what the attribute
2:43function does. What I'm actually going to
2:45do is I'm going
2:45to bring this calculation window back into
2:47context now. I'm just going to call this
2:50attr state. Okay
2:52and what we're going to do is wrap the
2:54state inside of the attribute function. So
2:57let's go
2:58ahead and just write state when we do that
2:59tableau auto complete. So when I just click
3:02that
3:02and then what I'm going to do is I'm
3:03actually going to highlight this is another
3:05hack you can do
3:06when you highlight a field then you double
3:08click the function. Tableau automatically
3:10puts the
3:10function around the thing you've
3:11highlighted. So that's a quick way of
3:13adding attribute to the
3:14whole entire state. Click apply and now we
3:16've got that attribute function here on the
3:18bottom left.
3:19I'm going to go ahead and drag that next to
3:21the state and you'll see essentially
3:25nothing happens.
3:25And this is fundamentally because of the
3:28level of detail that we're working with
3:30inside of this view. Because we are at the
3:33state level of detail the attribute
3:35function is working
3:36correctly. If I just go back to the
3:39description and we just click on this text
3:41here. When you
3:42click on these sometimes they're a little
3:44bit buggy but when you click on the name
3:45here in
3:46the calculation window this always works.
3:48Now returns the value of the given
3:49expression if it
3:50only has a single value for all the rows in
3:53the group. So it's working in this sense
3:55because
3:56everything only has a single value because
3:58the level of detail we're working at is
3:59state. So
4:00there's a one-to-one match for nearly
4:02everything for everything in this table in
4:03fact. Okay now
4:05the thing it will do if there's multiple
4:07values is return an asterisk. So the way we
4:09do that the way
4:10we create this scenario in our view is to
4:12just remove state. So I'm going to go ahead
4:14and remove
4:14state and leave the attribute only. Okay
4:17and you'll notice that now we get the aster
4:20isk that
4:20it was talking about earlier on. The other
4:23thing to note is that it's calling our
4:25calculation and
4:26aggregate. This agg notation lets you know
4:29that this is an aggregate. Okay so we've
4:31somehow managed
4:32to create an aggregate function around
4:34state names essentially. So it's somehow
4:37sort of creating an
4:38aggregation around these things and
4:39essentially returning just one value if it
4:41's the only value
4:42and an asterisk if there's multiple values
4:45and that's essentially what the attribute
4:47function
4:47does. It's pretty much it in a nutshell. If
4:50this is all you take away from the video
4:52then that's
4:52all you need to know. Of course there's a
4:54little bit more to this so we're going to
4:56try and dive
4:56into that in more detail. Let's try and do
4:58that. Okay so we now know what the
5:00attribute function
5:01does. Let's talk about when you might use
5:04this. Okay now if I just bring out a bar
5:06chart here
5:07let's bring cells onto columns to build out
5:10a bar chart and what you'll see here is
5:12that you get
5:12some nice bars here. If I just sort this by
5:16biggest to least you'll see that if I just
5:19undo
5:19that step actually if I just hover over
5:22this particular city Jacksonville you'll
5:25see that
5:25the attribute state has a star and if I go
5:28to the one above it California is in that
5:31place. Now
5:33what is quite common is when you're doing
5:35things like blending inside of Tableau you
5:38will often get
5:38a scenario where multiple items match to
5:41one value and in the visualization that you
5:43're asking Tableau
5:44to create it is not actually possible for
5:47it to aggregate that particular value down
5:50any more or
5:50any less and so it will just return a star
5:53in its place. Okay so in this case if I was
5:55to take
5:56attribute state away from here and put it
5:58on the tooltip you'll see that when I hover
6:01over
6:02what's this Columbus I get a star if I go
6:05to Detroit I get Michigan. Okay so this is
6:07actually
6:08a typical use case for the attribute
6:09function it's essentially telling you hey
6:11there's more detail
6:13underneath this data set. Now back in the
6:15old day you couldn't really solve this
6:17problem easily
6:18because it was really difficult to sort of
6:20create a way for people to drill in you
6:22probably have to
6:22have another chart somewhere else now we
6:25have charts and tooltips and you can just
6:27create a
6:27chart that shows okay yeah this particular
6:30one city exists in these three different
6:32states and
6:32so you can break that view down a little
6:35bit more but in the olden days you couldn't
6:37do that so this
6:38was actually a nice way of letting the user
6:40know there's more detail here they need to
6:41dig into so
6:42that's one use case for that. Now for the
6:44next use case this is the one that actually
6:46came up in the
6:472020.4 video where sometimes when you're
6:49working inside of a calculation you come up
6:51against this
6:52aggregation error okay I'm going to sort of
6:54forcibly create a similar scenario just by
6:56opening up the calculation window and we're
6:58basically going to create like a pretty
7:00pointless
7:00calculation but it should we should try and
7:03make it sort of valid if that makes sense.
7:05So I'm just
7:05going to call this sales in city okay and
7:09what I'm going to say if the city is null
7:12yeah so if is null
7:14city let's just type this in
7:18then null
7:24else sum of sales okay and and man I have
7:32written this in some sort of camel case I
7:34don't know
7:35anything about here so let's just use the
7:37suggestions that it's uh it's sort of
7:40giving
7:40us here I accidentally clicked else if in
7:42that case let's just type in else and you
7:44'll see if I
7:45just sort of um if I just sort of make this
7:47larger you'll see that I'm getting an error
7:51here okay and
7:51if I go down to the bottom here it says
7:53cannot mix aggregate and non-aggregate
7:56comparisons or results
7:57in if expressions okay so what's going on
8:00here okay let me sort of say this in plain
8:03English
8:03it's basically saying that over here I'm
8:05doing some aggregation I'm summing
8:07everything up
8:08and yet over here I'm basically just
8:11checking a single city so it's basically
8:13saying hey you
8:14can't simultaneously ask me to add
8:16everything up but then the beginning of the
8:18calculation asks me
8:20to compare with something that isn't added
8:22up okay so it wants me to aggregate the
8:24city in this
8:24particular case now the way I do that is
8:27coincidentally with the aggregate function
8:30okay
8:31and the reason it makes sense here is
8:32because when we then bring it into the view
8:34as long as
8:35we've got city on rows this calculation is
8:37still going to make sense so let's go ahead
8:40type in
8:40attr so this is the trick I used last time
8:43essentially just helped me clear out this
8:45error
8:46that I was getting and now you see the
8:48calculation is valid so if you're trying to
8:50write a calculation
8:51and you're getting into some aggregation
8:53issues and the thing you're trying to
8:55aggregate doesn't
8:56make sense to aggregate like a name or
8:57dimension or something like that that doesn
9:00't make sense
9:00because it's it's a sort of a grouping or
9:03an attribute of a dimension then this is
9:05what you
9:06do you essentially use the attribute
9:08function to sort of try and get past that
9:10error and now you
9:11can see the calculation is valid if I just
9:13go ahead and use that in my field I just
9:15say apply
9:16cells and city drag that next to that and
9:19you'll see it's exactly the same chart
9:21there is no issue
9:22here but if I was to remove the attribute
9:24item there then it would cause us an issue
9:27so that's
9:28pretty much the other use case so there's
9:30three use cases really like I showed you
9:32telling you
9:32there's multiple items inside of a
9:34particular grouping or particular level of
9:37detail the second
9:38is getting information into tooltips so
9:40when someone hovers over a bar chart or
9:43something
9:43like that where there is more detail in
9:45there then they can see that I'll just
9:48close this and I'll
9:49show you again this is still sort of the
9:51case here when I hover over Springfield
9:53here I get a star
9:54when I go to Jacksonville I get a star when
9:57I go to San Diego I only get California
9:58because that's
9:59the only item there and then the other way
10:02of doing this is of course in a calculation
10:05window
10:05where you have a little bit of a challenge
10:08so if I go in here cells and city you're
10:11trying to do
10:12an aggregation it's not letting you this is
10:14what's going on okay now we're not done yet
10:17because
10:18there's still more places you can see
10:19something called attribute okay what we
10:21have here is a very
10:22simple view and I'm going to actually work
10:23from here onwards I'm going to remove cells
10:25and city
10:26from the top row so we just got cells and
10:29city in context now the next thing I want
10:33to do is add
10:34something into this view and the thing I
10:36want to add is going to be at a higher
10:38level of detail
10:39than city okay so I'm going to add the
10:42country field let's go ahead here sorry it
10:46's a little
10:46bit further up this hierarchy I'm going to
10:49add country onto detail now when I do that
10:51it doesn't
10:52change anything meaningful in the
10:53visualization which is a good thing okay
10:56but watch what happens
10:57as I start to expand these items let's just
11:00add a region and now you'll notice that
11:02Springfield
11:03because it exists in multiple states can
11:06also exist in multiple regions okay so you
11:09can see
11:09here that this region is sort of west this
11:13is south this is east and this is central
11:17okay
11:18and so what's actually happening here is
11:20this our region field is controlling the
11:22level of detail in
11:23our visualization I know that because it's
11:25got this little icon here which is the same
11:28as this
11:28one here in the marks pane which is
11:30basically telling you that hey this item is
11:32actually
11:33affecting the lowest level of granularity
11:35in the visualization how do I get around
11:37that well this
11:38is how if I want to keep region in the view
11:40so that it can be used in a tool tip but I
11:42don't
11:43want it to affect the visualization what I
11:45can do is I can essentially turn this into
11:47an attribute
11:48so if I go in here click this drop down you
11:50'll see that one of the items here just
11:53after dimension is
11:54attribute and when I do that notice that it
11:57no longer affects my visualization so this
12:00is yet
12:00another use case for this particular
12:03capability it's going to do exactly the
12:05same thing so if I
12:06go back to Springfield you'll see that I
12:09now get a star for my regions okay
12:10essentially what this
12:12has done is it stopped region from
12:14controlling the way my visualization works
12:16it stopped it from
12:17sort of messing up the level of detail and
12:19actually that's a very deliberate trick
12:21that you can use if
12:22you just want to bring something into the
12:24view but you don't want it to interfere
12:25with your chart
12:26especially if it's the dimension you can
12:28just bring it in you can do this with
12:30virtually
12:30anything so if I go ahead here and get the
12:32product name if I put this on detail you
12:35notice it
12:35completely messes up my chart turn it into
12:38an attribute and boom my chart goes back to
12:40normal
12:41but now the product information is in the
12:44view the problem is here is that there's
12:46just so many
12:47products I'm always going to see a star
12:49here so it's really sort of pointless but
12:51if I need the
12:52product name in there this is going to work
12:55now if you're really advanced with Tableau
12:57and you'll get
12:58into use cases where having this
13:00information in the marks pane is necessary
13:02in order to achieve
13:03something else but that's for another video
13:04and that's for another day so that's pretty
13:06much the
13:07last use case you'll see for this okay
13:09limitations this is all we've got left to
13:11cover now there are
13:12some limitations with using the attribute
13:15function I've just reset the view here so
13:17we're back to a
13:18plain old data source I've cleared
13:19everything from the sheet and we've pretty
13:22much reset the work
13:22because if we've started from a fresh now
13:25one of the things I want to show you is to
13:26do with
13:27dynamic sorting what I'm going to do is I'm
13:30going to drag sales onto text then I'm
13:32going to drag
13:33subcategory onto rows this gives us a list
13:36of subcategories and essentially the values
13:38that go
13:39with them now I'm also going to create a
13:41calculated field I'll right click on the
13:43field create
13:43calculated field and what I'll do for this
13:46one is I will just set this to an attribute
13:50as we did
13:51before so I'll just wrap this in the
13:53attribute this time I'll type it out just
13:55just to show that
13:56I'm a lazy person and I'll call this ATTR
13:58sub cat okay so now we're pretty much ready
14:01to go that's
14:02all good hit apply perfect it's now gone
14:04over here to the bottom left hand side now
14:07if I drag that
14:09next to the subcategory you'll see that
14:11basically it says the same thing but when
14:13it comes to sorting
14:14it's a slightly different story if I go to
14:16the first one here and I click sort you'll
14:18see that
14:19I have the full capability of sorting I can
14:22do all the sorting I want including any
14:24dynamic sorting
14:26based on a field so you can see that the
14:28sort changes depending on what I'm actually
14:30counting
14:31here the minimum the maximum each time this
14:33is changing this is called dynamic sorting
14:36however if I then go to my second one let
14:38me just clear the sort when I go over here
14:42and I go to
14:42sort you'll see that I can only really sort
14:45it by the manual formats there's no dynamic
14:48sorting
14:49available to me here because I'm using an
14:51attribute function around the dimension so
14:53just a word of caution you lose this
14:55ability when you're using an attribute
14:57function so just be
14:58aware of that if you need to sort with an
15:00attribute you lose a lot of capability okay
15:02the next one is
15:03slightly easier to show you okay let's just
15:06open up a new sheet and what I want to do
15:08here is I
15:09want to create a grand total so what I'm
15:11going to do is I'm going to drag sales put
15:13it onto text again
15:14order date onto rows actually I want to put
15:17this on columns not rows and then I want to
15:19bring the
15:19subcategory here again okay and here we
15:22have a list of values now the use case here
15:25is that if
15:26you've used an attribute inside of a
15:28calculation what could happen sometimes is
15:30that the grand
15:31total won't render correctly okay let me
15:33show you like a forced example it's not
15:36quite the same as
15:37sort of a standard example you'd bump along
15:39because you would never write a calculation
15:41like this but
15:42I've just got to try and sort of recreate
15:43the scenario here so let's just close the
15:46side window
15:46because I now know what the attribute
15:48function is and I'm just going to sort of
15:50forcibly type
15:50something so I'm going to say if attribute
15:56not the subcategory if attribute actually
16:02yeah we can
16:02actually reuse that if attribute sub
16:04category let's use that let's use the same
16:06field we've just
16:07brought in I was going to do something else
16:09but we've already created a function that
16:11does that
16:11equals and let's say we're checking for
16:14something here a simple string let's say
16:18art
16:20then some of cells okay and I forgot my
16:24brackets so nothing
16:26automatically formatted let's just close
16:29that off and capital s for cells and now it
16:33gets going man
16:34I cannot spell here we go then some cells
16:36uh end okay we don't need to do an else
16:38because there's
16:39nothing else coming after that it just
16:40assumes that it's null if there's nothing
16:42after that okay
16:43so if attribute equals r then some of cells
16:45end okay so we'll call this attribute cells
16:48just to
16:49show that it's only going to show uh if the
16:51art attribute is basically the item so I'm
16:54going to
16:54hit apply hit okay and what we're going to
16:57do is we're going to remove this cells
16:59value from the
17:00view and we are going to put this on text
17:03instead so so far so good it's doing
17:06exactly as we
17:07expected we're getting the individual
17:09values for art and for nothing else however
17:12if I then go and
17:14enable a grand total a column grand total
17:16in this case you'll see the total doesn't
17:19work correctly
17:20and so this is another use case for
17:22attribute or limitation is that grand total
17:25sometimes won't
17:26render correctly now the fix for this is to
17:28use a min or max but then again might not
17:30render the
17:31right values so just be very careful about
17:33how you sort of solve this this really
17:35typically points to
17:37where the data prep issue you probably want
17:39to solve this problem further back in the
17:41stack as
17:41it were as the data comes in maybe doing
17:43something like order of operations to get
17:45around this problem
17:46LODs exist now as well so you can actually
17:48do this sort of more dynamically using that
17:51anyway
17:52but just something to be aware of and then
17:54the last thing is to do with performance I
17:56'm just
17:56going to build a very simple list actually
17:59maybe what I'll do is I will take a
18:01category
18:02and subcategory and put that next to each
18:05other okay so we've got a bunch of subc
18:08ategories that
18:09are part of categories now when we use the
18:11attribute function in this particular use
18:14case
18:14I'm not actually going to write anything I
18:15'm just going to stay on the screen what the
18:17attribute
18:18function has to do when it looks at each
18:20grouping here by grouping I'm talking about
18:22the categories
18:23is that for each category it has to go and
18:25look at what's inside of the subcategory so
18:29it will
18:29start with bookcases and then realize that
18:31bookcases and tables are not the same thing
18:33and so by doing that what that means is it
18:36has to load the entire data set in there
18:38okay now
18:39one of the things is that Tableau is
18:40actually very efficient and it sort of when
18:42you use something
18:43like an extract it pre-indexes the data and
18:45it understands a little bit about what's
18:47going on
18:48and so if you just need the smallest or the
18:51largest value in a particular table it's
18:54actually not a
18:55good idea to use attribute if the result
18:57would be achieved the same way if you used
19:00min or max
19:01okay because min or max will be much faster
19:03at finding what's first and last than
19:05something like
19:06an attribute so it's something to be aware
19:08of sometimes you're using attribute to
19:10solve a
19:10problem in the visualization when in fact
19:12you don't actually even need the capability
19:14that
19:15attribute is offering you it's just getting
19:17past the viz like I showed you earlier so
19:19in those
19:19cases you might be better off using using
19:21something like min and max to get around
19:23the problem rather
19:24than using something like attribute so it's
19:27a bit of a weird one to sort of show you
19:28and I need to
19:29think sort of more creatively in terms of
19:31examples maybe if you've got an example
19:33post it in the
19:34comments below and other people can benefit
19:36from that but that's pretty much it for
19:37limitations
19:38those are the three kind of key ones those
19:40are the ones that tabla also document on
19:41their help pages
19:42and that are worth highlighting so that's
19:44pretty much it that is the attribute
19:46function in a
19:46nutshell and there's definitely more we
19:48could have dived into but fundamentally
19:50this is all you
19:50really need to understand in order to use
19:53this day to day the first one is just using
19:56it to tell you
19:56if there's more detail underneath the thing
19:59you're analyzing it will show you a star if
20:01there is and
20:02it won't if there isn't it will show you
20:04the lowest level item in that particular
20:06group the other use
20:07cases inside of a calculation and the final
20:09use case is in this view where you need to
20:12change the
20:12level of detail without removing
20:14information from the marks pane the rows
20:16columns or anything else
20:18okay that's pretty much it in a nutshell if
20:19you've enjoyed this you know what to do if
20:21you haven't
20:22let me know in the comments below and I'll
20:23catch you in the next video
Tableau | Data visualisation | Analytics