0:00Hey it's Tim here in 2021.2. Tableau have
0:03enabled explained data for viewers but they
0:05've also given
0:06the explained data interface a massive rev
0:08amp and they've given authors additional
0:10controls as to
0:11how explained data works for the viewers
0:13who can now access the feature. The thing
0:15is though it's
0:16not enabled by default so in this video I'm
0:18going to show you the new interface and how
0:20to set it
0:21up so viewers can use this in their own
0:23exploration of data. Right let's get stuck
0:25in I'm here in
0:26Tableau 2021.2 I'm actually just going to
0:29open the sample superstore workbook this is
0:31a workbook that
0:32everyone has so we can just go ahead and
0:34open it and once this is open I'm just
0:36going to go through
0:37essentially what's changed with explained
0:39data once this opens up and here we are we
0:41have the
0:41executive overview this is a very standard
0:44sort of sample workbook it's pretty much
0:46the only sample
0:47workbook you should use to compare
0:49performance on Tableau server and Tableau
0:51desktop but anyway we
0:52want to go to the product level breakdown
0:55because here we can actually get some
0:57useful analysis for
0:58our data and I know this is the same for
1:01everyone so it's a good place to go now if
1:03I go over here
1:04to the corporate tab you'll see that I have
1:06two data points that are a little bit sort
1:08of
1:09different compared to the other data sets
1:11so as a viewer what I might want to do is
1:13you know
1:14understand why that is what's causing the
1:16behavior to be so different in these
1:17particular data points
1:18and so in the past what you've been able to
1:21do is to just click on the data point and
1:22of course if
1:23there's any actions or filters that of
1:25course changes the visualization but the
1:27key thing here
1:28is the explained data has previously been
1:30in the past sort of not been available for
1:32viewers that's
1:33this option here it's actually been
1:34available for authors so that they can
1:36understand what's going
1:37on and it can sort of help them you know
1:39understand where to go next with their
1:41analysis but now this
1:42option is available for viewers and so when
1:45we click on this little option here we'll
1:47get this
1:48new interface for explained data now you'll
1:50notice a couple of things previously it
1:52used to be a pop
1:53up window where it opened up in the middle
1:54of the window and you could see the
1:56analysis and drag
1:57things around this time around we've got a
1:59new interface item for tableau and I think
2:01this is
2:02actually a bigger deal than it is because
2:04the fact that we've got this whole pane on
2:06the right hand
2:06side suggests that this won't be the only
2:08thing that will live on this right hand
2:10side part of the
2:11pane but nonetheless explained data is the
2:14first thing that we see here and we can of
2:17course see
2:17the different analysis that it's pulled out
2:19now the way explained data works is it does
2:21some
2:21statistical analysis of the fields and it
2:24tries to understand hey what's causing this
2:26particular
2:27data point to behave this way and it does
2:29that by looking at all the fields pulling
2:32out the pertinent
2:33measures that it thinks are important the
2:34person dimension that it thinks is skewing
2:36things and it
2:37tries to summarize that for you in a
2:39detailed view so if we have a look at this
2:40we've got this
2:41explained mark and for the record the
2:43explained mark is always the mark you
2:45selected so this
2:46explained mark is this data point here if I
2:48selected a range of marks I think it would
2:50work
2:50over a broader range but generally speaking
2:52it works better on just one mark because
2:54then you're
2:54sort of focusing in on the question now um
2:58the office supplies hoover stove red
3:00corporate
3:01appliances essentially what's explaining
3:03this particular data point um and then
3:05there's three
3:06values here that's pulled out so profit
3:08sales and profit ratio okay so you're
3:09saying there's a higher
3:10than expected profit and then when you
3:12click on this arrow here on the right hand
3:14side it actually
3:14goes to a new page in this particular pane
3:17and it runs a little query in essence it's
3:19actually
3:20doing some analytics for you the fact that
3:21it runs that query means it's actually quer
3:23ying the data
3:24itself to try and give you the analysis
3:26that it thinks is pertinent and you'll see
3:28here that
3:28there's two key things that are important
3:30so you've got the mark attributes okay so
3:33record
3:33level values and their aggregates in the
3:35data source sample superstore may be
3:37contributing
3:38the value of sum of profit so if I then
3:41open this explanation it says the average
3:43profit here is
3:442179 so basically the average profit for
3:47this uh bunch of items is rather high and
3:50if you notice
3:51that it actually sort of drew a
3:53visualization for me there so it's really
3:55taking me deep into
3:56the analysis and now we're three levels in
3:58deep we've understand look the mark has
4:00this sort of
4:01strange behavior the average profit is
4:03super high and you can actually the really
4:05cool thing is this
4:06is actually a chart you can actually
4:08interact with it and just it's like you're
4:09in a dashboard
4:10and you can actually interact with it and
4:12you can then start to see how this might be
4:14skewing your
4:15data okay the other really cool thing is
4:17this really small icon here where you can
4:19actually
4:19pop this chart out so if you want to bring
4:21out this chart if you click on that icon it
4:24actually
4:24creates the chart for you I don't know what
4:26it's going to do here I haven't actually
4:27tried this
4:28myself I think it creates a new sheet and
4:30there we go it keeps the pane open on the
4:32right hand side
4:33and now we have this exact same chart now
4:35available on the main screen and actually
4:38this
4:38is analysis that you can take on the only
4:40downer here is it's put this label over the
4:42data point
4:43it's sort of not so smart this little label
4:45supposed to be pointed to this data point
4:47and so it just explains that anomaly really
4:49well this is nice if you need to just print
4:51this out
4:51very quickly put this in a pdf or you've
4:53got some edge case data points you've done
4:55some dashboards
4:56and then there's three data points you can
4:58quickly use explain data hopefully to get
5:00at these insights
5:01much much more quickly okay now the key
5:03thing is it's opened up a new view and our
5:06selection has
5:07changed so one thing you'll notice is that
5:09it's telling us just here that your
5:10selection has
5:11changed essentially it's noticed that what
5:13I've picked in my view and generally
5:15speaking if we're
5:16talking about views I'm talking about what
5:18I've picked in this sort of box here in the
5:20space that
5:21I build my visualization it's telling me
5:23that that's changed because that's changed
5:25the analysis
5:27that it's doing which is over here in this
5:29explain data window which I'll just go
5:31behind my face here
5:32is actually no longer accurate okay so it's
5:34basically just telling you hey your
5:36perspective
5:36has changed you might want to reselect
5:38things okay but I don't want to actually do
5:40that what I want
5:41to do is go back to the analysis that I was
5:43doing which is over in product and you can
5:45see that when
5:45I go back here now this little message that
5:48was here is gone so it has some contextual
5:50awareness
5:51of what's going on I'd love a little back
5:53button here in the explain data to take me
5:55back to
5:56where I was doing the analysis but
5:57nevertheless it does remember where you are
5:59and it sends you
6:00back okay so that's the first thing we've
6:02only just done the first bit of analysis
6:04here on mark
6:05attributes the other thing if I just
6:07collapse this in we'll see that we have
6:09relevant measures here so
6:10the measures may be contributing to the
6:13higher sum of profit so basically it's
6:15looking at the average
6:16profit and then it's basically looking at
6:18the sales forecast and the sales so if we
6:20open up the sales
6:21forecast the average self-forecast is
6:23increasing the expected profit of the
6:25explained mark now
6:26Tableau is just doing what it can to
6:28understand your measures so sometimes this
6:31analysis doesn't
6:32sort of quite hit the mark for me because a
6:34sales forecast is a sales forecast how can
6:36that be
6:37increasing the profit it can't so in this
6:39analysis it's just basically looking at the
6:41forecast it
6:42doesn't understand that the sales forecast
6:44and the sales are two different things so
6:45in this case I
6:46can sort of disregard the sales forecast
6:49but I'll show you how to solve that
6:50properly in a second
6:52and if I go to sales here this is the
6:53actual field we're interested in the
6:55average sales is increasing
6:57the expected profit of the explained mark
6:58so in this case I think this is accurate
7:00because this
7:01is not our forecast this is our actual
7:03sales value so in this case you can see yet
7:05another chart and
7:06if I hover over it it gives me the detail
7:08that I need to see there and I can actually
7:10start to work
7:11with it more clearly now you'll notice that
7:13I said before that the sales forecast wasn
7:15't pertinent
7:16so you can actually remove this from the
7:17analysis so let me go I'm actually going to
7:19need to move
7:20my face up here to the top and what I
7:22essentially need to do is you see this
7:24little gear icon when
7:25I click that I get this amazing pop out and
7:27this pop out does a couple of things number
7:29one it
7:30shows me what fields are being used in the
7:32analysis and number two it shows what
7:34explanation types are
7:35actually being computed in this data set
7:37and most importantly it also gives me a
7:40place where I can
7:41actually enable this for viewers so if you
7:43don't tick this box viewers won't be able
7:45to use explain
7:46data when they're doing their analysis so
7:48you can keep this as a creator author only
7:51feature
7:51or if you tick this box it's not ticked by
7:53default you'll actually enable this for
7:55viewers to do their
7:56own insight now when you do that as you've
7:58just seen you know it pulled up sales
8:01forecast as
8:02something that you know is is important
8:04when I know it's not important so you need
8:06to be able to
8:06sort of go in here and curate these things
8:08you might want to change for example some
8:11of the
8:11analysis it's doing you might have a data
8:13set that doesn't lend itself well to
8:15extreme values because
8:16it's a you know in an edge case data set
8:18you might have some sort of analysis that
8:20doesn't make sense
8:21of number of records doesn't make sense
8:23because you're dealing with really large
8:25data sets where
8:26things move around a lot and so although
8:27tableau thinks it's relevant it might not
8:29actually be
8:30relevant because you understand the beta
8:32data better than it does so these are all
8:34things you
8:34can do and when you tick this box allow
8:36explain data to be used in the workbook
8:39when viewed online
8:40you get additional options so you allow all
8:42users to see extreme values explanations
8:44with record
8:44level data and do not show extreme value
8:47explanations so by default this is on do
8:49not
8:50show extreme value definitions essentially
8:52by default tableau is not going to start
8:55talking
8:55about what's happening at the extremes of
8:57your data but it will still show you as a
8:59creator
8:59you'll still be able to do that but if you
9:01want to enable that you can of course tick
9:02that radio
9:03button and it gives you a warning extreme
9:05value explanation display record level
9:07information in
9:08underlying data do not enable extreme
9:10values if your workbook uses sensitive data
9:13so essentially
9:14this is a security thing that you need to
9:16be aware of because if you enable explain
9:18data and then you
9:19don't give people sort of the ability to
9:21drill down to that data and then through
9:22explain data
9:23they're actually able to get to it that's
9:25not going to be a great look so again it's
9:28off by
9:28default so you can't really make this
9:30mistake but just be aware of this warning
9:32that it's really
9:33really important okay so nonetheless you do
9:35get a lot of controls here now before when
9:38i came to
9:38this fields page you'll notice that it didn
9:40't show up i had a little bug and it's now
9:42loaded you can
9:43actually change what's being analyzed so i
9:44can actually go down here to some of sales
9:46forecast
9:47and you'll see that it's an automatic i can
9:49set this to never include this in analysis
9:51because
9:51the forecast is exactly that in this case
9:54it's actually in a forecast based on a
9:56parameter
9:57so it doesn't make any sense to have that
9:58in there essentially it's just a parameter
10:00that lets you
10:00choose the percentage and that changes the
10:03forecast based on ourselves so it doesn't
10:05make
10:05sense to include that in this analysis so
10:07if you click never include you click okay
10:09now that will
10:11never be included in my analysis and so
10:13although it's in here now if i was to click
10:15out again and
10:16i was to reselect things okay so if i was
10:19to go in here and then do this again and
10:21then just do this
10:22and you should see that it will think about
10:25it again and if we go back into profit and
10:29we look
10:29at the sales analysis that it did we should
10:32now notice that the sales forecast is no
10:35longer there
10:36and you can see that that's worked exactly
10:38as described we only have sales in here so
10:40you do
10:40get control but again you've got to spend
10:42the time you've got to invest some energy
10:44into making that
10:45work okay i'm going to move my face back
10:47down here just where i'm used to it being
10:49and i'm just going
10:49to go back so that's just a deep dive into
10:51one of the analysis that it's doing you can
10:54see here
10:54that's done profit sales and profit ratio
10:56and again each of these analysis it's going
10:59to choose
10:59different mark types and different things
11:01each time and you've got to really sort of
11:03you know
11:03play around with your data and explore it
11:05but i think this is a nice addition because
11:07it gives
11:07people the ability to just go a bit beyond
11:09what they've been given in terms of
11:11reporting and it
11:12can also help people sort of articulate
11:14better what they'd like to see maybe there
11:16's a there's
11:17an angle in one of these extreme values and
11:19maybe there's an angle in another part of
11:20the chart that
11:21you know works a lot better now again the
11:23analysis that you get there is really
11:25really powerful
11:26but tableau is always making a guess as to
11:29which fields it's using so you can see here
11:3216 of 35 fields now when you click on that
11:34little blue text if you're wondering where
11:36i went i'm just
11:36down here on the bottom right hand side and
11:39when you click on that it actually shows
11:40you what was
11:41used and what was excluded in that analysis
11:43so every time you tick on something or you
11:45select
11:45something it's going to look at it and try
11:47and guess what it should include and what
11:49it shouldn't
11:50include let's just try that with another
11:53data point so let's just go in here and if
11:55we go to
11:56another chart let's go to this customers
11:57tab i want to go to completely different
11:59visualization
12:00okay and let's select um i'm going to
12:02select sales here at the top and you'll
12:05notice this time i've
12:06selected a grouping i've selected a
12:08dimension and you'll notice that you see
12:10you don't get us data
12:11with this particular feature that's
12:13something to be aware of you know it only
12:15works on data point
12:16so now if i go and select central you'll
12:18see that it turns up and now when i select
12:20that again it
12:21goes off and does another analysis and you
12:23'll see that i get a lot more this time i
12:25actually get
12:26a lot more of a story going on and there's
12:28a lot going on here and again this might
12:30just need a
12:31little bit of pruning as an author to make
12:33sure this makes sense but essentially if
12:35you go into
12:36any of these marks you'll start to
12:38understand that it's actually doing a lot
12:40of analysis behind the
12:41scenes and this time around you'll see
12:43instead of 16 it's actually given us 18 of
12:4635 fields so every
12:47time it's doing an analysis it's trying to
12:49make an educator guess what fields actually
12:52matter and
12:52it's using statistical methods that frankly
12:55i don't fully understand but um you can
12:56actually
12:57click on this learn more link and it opens
12:59up a tab i'll just bring this into the
13:00window here
13:01it's opened up in my other screen and you
13:03'll see here how it works and so you can
13:05actually go in
13:06and read more about this sort of type of
13:08analysis and breakdown that it does but
13:10essentially it's
13:11it's actually sort of well thought through
13:13now like all of these features that you use
13:15sort of
13:15machine learning and ai it's useful just to
13:18bear in context that you really need to
13:20sort of understand
13:21where these features are coming from um
13:23they might drive your analysis they might
13:26inform where you go
13:27next but you still have to apply a little
13:29bit of rigor and thought in your own work
13:31as a creator
13:32to say is that does that actually make
13:34sense in this analysis or with this data
13:36set okay so that's
13:37the explain data feature set i think it's a
13:39really cool addition the fact that it's
13:42available for
13:42viewers is a really big thing as well
13:45because i think for a long time tablo has
13:47been a very
13:48creator-led product and it's great to have
13:50some of this power come to viewers in a way
13:52that's so
13:53much easier for them to instigate without
13:55ever having to write a calculation or set
13:57anything up
13:58and it's leading on the work that authors
14:00are doing as well because of course you
14:02know authors
14:03build these data sets and and sort of make
14:05these available to people so that's a
14:07really really quick
14:09summary of the new explained data features
14:11in 2021.1 i think this is actually a bigger
14:13deal than
14:14than i can sort of emphasize in this video
14:17because um a couple of things you know just
14:20just the simple
14:21fact that they're willing to sort of bring
14:23this whole panel here and here and sort of
14:25bring it in
14:26on the right hand side really suggests that
14:28tablo is moving in a direction where they
14:30want to help
14:30the user understand things and this right
14:33hand side panel is going to be where that
14:36context is
14:37conveyed it's going to be where that story
14:40is told and explained data is the first
14:42thing here
14:43our state is sort of different because it's
14:45more of a an analysis that starts with the
14:47question
14:48but explained data somewhere where you know
14:50some explanation is given context of our
14:52data seems to
14:53me like he's going to come here now the
14:55other thing i would actually love is if
14:57this pane was made
14:58available to authors so it's great that
15:00explained data can have this space i'd love
15:02to put filters
15:03in here i'd love to put controls in here so
15:05fingers crossed they've done half the work
15:08they've put
15:08their own feature in there i'd love to be
15:10able to put my own features in there maybe
15:12add a little
15:13filter section in there just to make my
15:15dashboard a little bit more user friendly
15:17and easy to use
15:19but that's pretty much explained data in a
15:20nutshell it's not a new feature but it has
15:22been made available to viewers and they
15:24have given it a revised interface so it's
15:27no longer
15:27a pop-up and it's actually very prominent
15:29and now everyone can use it along with
15:31their license so
15:32that's really good to see thanks for
15:34watching if you enjoyed this video you know
15:35what to do
15:36otherwise i'll catch you in the next video
15:39about 2021.2
Tableau | Data visualisation | Analytics | AI & ML