Hexbins & Mapping in Tableau: Tutorial HD with CC June 2016

0:00Hey, Tim here and today I'm going to be

0:02showing you how to create hex spins in

0:05Tableau. I was reading recently a paper

0:09from the Tableau research website on

0:12hex spinning and the paper basically talks

0:14about the distortion that takes

0:16place when you do hex spinning in a

0:19geographical context because of course

0:22the world isn't flat and so when you

0:25project hexagons or any sort of shape

0:28mapping onto a map then actually you're

0:31distorting the information so the paper

0:33covers this is a quite a geeky paper so if

0:35you're into that kind of stuff

0:37definitely head there but it got me

0:39thinking about hex spins in Tableau and

0:41so I wanted to show you today how to do

0:44that for data that might have an X and

0:46Y axis or mapping data so a lat and a long.

0:52The data I'm going to be using

0:53today is actually from this site here it's

0:55just dummy data I've created a

0:58basic schema here so you have customer

1:01details you have the X and Y axis

1:04and then you have latitude and longitude

1:06and just for fun I've added in some

1:08country data in there as well just so we

1:11can very easily distinguish between two

1:14regions. The data I'll be using I'll be

1:17putting that into 9.3.3. Hex spinning

1:21is supported from Tableau version 9 onwards

1:24so that's a really important

1:25requirement if you want to follow along

1:27make sure using version 9 onwards. Okay

1:30so I'm going to connect to my mock data and

1:34as I showed you we have these fields

1:38here so you have some customer data some

1:41country data some emails and then you

1:43have the four fields we're going to be

1:45looking at when we do hex binning. Now

1:48hex binning itself is just like binning in

1:51Tableau so if you wanted to say bin

1:53all your measures you can just right click

1:57on the option go to create and

1:59then select bins and it gives you some

2:02options here about how big those bins

2:06are. Now hex binning is a very similar

2:09concept except for what you're really

2:11doing is you're doing it on a two-

2:13dimensional plane so you're

2:14basically taking items that are near each

2:17other and you're putting them into

2:19a bucket and having one data point to

2:22represent all those sort of neighboring

2:26data points and the function that you use

2:28to do that is called hex bin so when

2:32you type in hex you get these two functions

2:36here hex bin X and hex bin Y

2:39and you need both of them when you create

2:41your hex bin calculations you

2:43need one for the x-axis and one for the y-

2:46axis and if you're doing this for

2:48mapping you need one for the latitude and

2:50one for the longitude so first I'll

2:52start by showing you how to do this for the

2:54x and y-axis so first of all I'm

2:57just going to hit enter to capture the

3:00first hex bin function and then here I'm

3:03just going to type X which is one of my

3:06fields and then Y and that's it we've

3:10created the hex bin X calculation so I'll

3:13give this the value hex bin X

3:15because I'm lazy I'm going to copy this I'm

3:18going to click OK I'm going to

3:20create another calculation I'm going to

3:22paste that in there I'm going to call

3:24this hex bin Y you can see I've just

3:30changed that function there and click

3:33OK so now we have these two new

3:35calculations now step number two the

3:39really important bit is to make sure that

3:41you move these from the measures

3:43pane up to your dimensions pane so I'm just

3:47going to select both of these I'm

3:49going to move them up okay and now they're

3:51there and now we can actually

3:54start to put them into our visualization so

3:56I'm just going to bring out hex bin X

3:58and I'm going to bring out hex bin Y now

4:02immediately this doesn't look like what

4:05I was trying to create I can sort of see

4:07some things happening but it I can't see

4:10any sort of shapes and that's actually

4:12because I've dragged out dimensions onto

4:15the visualization and although we wanted to

4:17do that in the first place when we

4:19visualize them we want it to be done on a

4:22continuous scale so we need to

4:24actually change these to continuous items

4:29okay and as you can see we now get

4:33something that looks fairly close to what

4:35we're trying to achieve you seem to

4:37have data points grouped in a very

4:39organized way you can even see gaps in

4:42the data where there isn't data with

4:44neighboring sort of cohorts and if I

4:48very quickly plot the X and Y axis I can

4:50show you this is actually working

4:53because if I do this and I bring in the

4:56lowest level of detail which I think is

4:59I think it's going to be first name and

5:03custom ID and also those three put that

5:08on detail then that's actually what you get

5:11when you have the full level of

5:13detail so the hex binning is actually

5:16working and so this this is working but

5:20it still doesn't quite look like what we

5:22want to achieve yet there seems to be a

5:24very high density of binning going on so

5:28how do we address that okay well it

5:30turns out we have to edit the calculations

5:34that we created in the

5:35first place and to do that I'm actually

5:37going to create a parameter okay the

5:40parameter will allow us to change the

5:42valley that we're going to use you don't

5:44have to use a parameter but it gives us a

5:46little bit of flexibility and it will

5:48allow me to demonstrate to you how to do

5:50this on the fly so I'm going to call

5:53this parameter density density okay and I'm

5:59going to give it a range I'm going

6:02to give it a range of zero to a maximum of

6:06about ten and I'm going to make the

6:09step sizes quite frequent so I'm going to

6:14do not point zero two and I'm going

6:17to click OK and if I just show the

6:21parameter control you can see the

6:24density is one I'm going to leave that just

6:26there for now and now we need to

6:29adjust our calculation so if I click on

6:33edit and basically what we want to do is

6:35we want to basically multiply these X and Y

6:42values and then divide by a value

6:45so you want to increase the scope the

6:48tablet looks at the data at but then

6:52reduce the values that it doesn't skew the

6:55data probably not explaining that

6:57too well but let me just type the

6:59calculation in and you'll see where I'm

7:01heading so I'm just going to multiply this

7:04by the density I'm going to

7:07multiply Y by the density as well and then

7:12I'm going to divide the whole thing

7:14by density okay and that's hex bin X and we

7:18can actually just copy this part of

7:20the calculation click OK and then paste

7:31that here okay so you've seen me change

7:38these two calculations and yet nothing has

7:41changed now remember the density

7:43value is set to one so if I look at this

7:46value well of course X multiplied by 1 Y

7:50multiplied by 1 divided by 1 is going to

7:53give me the same value as I had before

7:55but the cool thing now is if I change the

7:57density for let's say go down a

8:00little bit you can start to see what's

8:02happening the groups are getting larger

8:05I'm getting less and less circles to the

8:09point where if I was just to take this

8:11down manually to not point zero one two I

8:15have much much much less space going

8:18on there okay so if I just bump up the size

8:22you can see and just as proof that

8:24this is working I'm actually going to bring

8:26in the number of records filled

8:28here and I'm going to put that on label

8:30okay and I'm going to make sure that

8:32this is centered like so so you can see it

8:36inside of the circle and I'm going to

8:39bring the size down a little bit so you can

8:42see what's going on as these get

8:44bigger or smaller let's go on let's make it

8:46a bit bigger actually okay so watch

8:49what happens at the moment we have roughly

8:53350 to 70 items per shape you

8:56can see on the fringes we have less density

9:00and in the middle you have

9:01around 380 in some cases 400 so if I bump

9:06this number up you can see the

9:09number of items in the show squares getting

9:12smaller and if I make that

9:15really small then you can see the text just

9:17about there we're now into the 50s

9:19so this is actually working and this is hex

9:22spinning working on the dashboard

9:24now the last step is actually making these

9:28shapes sort of interesting so in

9:31hex spinning it's actually named after a

9:34hexagon and right now I'm using a circle

9:37now the reason hex bins are used or hex

9:39agons are used is because they're

9:41actually the most compact way of having

9:44shapes side by side if you look at a

9:46honeycomb it's even nature's choice for

9:50having a lot of sort of shapes

9:52compacted into a very small sort of surface

9:55area and you know space so if I

9:59go to shapes here I've actually got my own

10:03hex spinning shapes I comment a bit

10:06crazy with these I actually have quite a

10:08few and I'll be attaching these to the

10:10blog post so if you're watching this on

10:12YouTube hit the link at the bottom in

10:14the comments and it will take you to the

10:16blog post and you can download these

10:17shapes and I'm gonna click apply and so

10:20here we have a very basic hexagon and if

10:24we increase this you can start to see what

10:27's going on and I can even just

10:29maybe bring the width in a little bit bump

10:33up the size and you start to get

10:37that iconic sort of grid that you come to

10:40expect with hexagons where everything

10:42is sort of very nicely and mapping or

10:45fitting side by side and obviously it

10:49looks like we're zooming in it's a bit

10:50psychedelic actually and you don't have

10:53to use a particular shape you can just use

10:55circles and if we make this a

10:58larger as you make this bigger you can

11:03obviously start to see that it just gets

11:06a little bit too busy but it works okay so

11:09we've done one type of hex binning now

11:12I'm gonna do exactly the same thing but for

11:15a map okay so I'm gonna go over to

11:17sheet three actually so this is what it

11:20would look like without binning and this

11:22is what it looks like now now for mapping

11:25we normally use a latitude and

11:28the longitude so let's have a look and see

11:30what that looks like on a map so

11:32let's just bring those two in I never seem

11:35to get these the right way around

11:36and at the moment all our data has an

11:39average around Italy but if we bring in

11:42again the customer level of detail you

11:47start to see we have a very big global

11:50data set now when I created this sample

11:52data I only actually brought in the

11:55dimensions for country if it was France and

11:57the United Kingdom so let's filter

11:59that down and now we've got our data set

12:03just for France and just for England and

12:05here we want to apply some hex binning so

12:08when we do hex binning on a map it's

12:11sort of different because we're still using

12:13the same technique but we need to

12:15actually apply it to some sort of X and Y

12:17axis and of course we don't have that

12:19as we'd expect to see it and that's where

12:21the latitude and longitude come

12:24and you see these are basically the X and Y

12:26values on a map so I'm actually

12:29gonna be a little bit lazy and I'm going to

12:34duplicate these okay and I'm gonna

12:40edit this I'm gonna rename this hex X and I

12:43'm gonna name this map okay and I'm

12:49gonna start with the longitude and latitude

13:00okay click apply and then I'm

13:04gonna do the same for this and as we do

13:13this click apply we now have two fields

13:21so let me just rename this correctly so

13:24that we have a consistent naming okay

13:29perfect okay so now what I'm gonna do is I

13:33'm actually gonna duplicate this so we

13:36have a reference point and in order to

13:40actually make this work we still need to

13:42tell tableau that this is a latitude and

13:44longitude so I'm gonna just change the

13:46data type here and give this a geographical

13:49role so that one latitude

13:51again geographical wire for that one long

13:56itude okay and then I'm gonna clear

13:59this map again I'm gonna bring in my eye

14:05and I'm gonna bring in my X and if I

14:11rotate these a little bit you start to see

14:13okay got some interesting thing

14:15going on here it is binning but first and

14:18foremost I want to bring in just

14:20France and the UK focus it in focus us in a

14:25little bit and the density still

14:29not quite right so what I need to do is two

14:32fold number one I need to bump up the

14:36density okay and you can see very clearly I

14:39've obviously made a mistake

14:40because all the data points are around

14:44Africa and obviously France and the UK

14:46are not around Africa so I've put these the

14:50wrong way round in fact not only

14:53about put them the end right now around I

14:56have also given these the wrong

14:59assignment very easy mistake very easily

15:02fixed okay so now if I put that there

15:07that we have the right lat and long for the

15:13shapes and now when I increase the

15:17density you can see very easily that I'm

15:20now binning within those countries and

15:22just to make it easy to see I'll put

15:24country on color by control dragging

15:27placing that on color and then one thing we

15:31can do is put number of records on

15:34size to show us where is the density of our

15:40customers and we can even go as far

15:43as saying okay now that we've done that let

15:46's adjust the range of this so that

15:49we can see very easily that we do have data

15:53spread across the countries but a

15:56smaller customer bases are in the outer

16:00regions of each country and the bigger

16:04customer bases are obviously centrally

16:07located and you can sort of play around

16:10with this until you get the optimum sort of

16:12distribution and again you can put

16:15instead of country you can put number of

16:17records on color and you can basically

16:20use different techniques here to try and

16:23make your data a little bit clearer and

16:26a little bit easier to decipher I'm just

16:29gonna give this a dark hello there so

16:31you can very easily see what's going on so

16:34that's how we do hex binning on a map

16:37and also an X and Y axis where you might be

16:42using scatter plots let me know what

16:45you think drop some comments in the feed

16:48below I'm gonna attach the workbook that

16:51I've used as is right now so you can

16:53obviously take this apart have a go

16:55yourself and I'd love to hear your feedback

16:57on the video if you've got some

17:00tips let me know and yeah I'll see you soon