Pictures from the Gallery 9: Innovative Graph Builder Views
A picture is said to be worth a thousand words, and the visuals that can be created in JMP Graph Builder can be considered fine works of art in their ability to convey compelling information to the viewer. This journal presentation features how to build popular and captivating graph views using JMP Graph Builder.
Based on the popular Pictures from the Gallery journals, the Gallery 9 presentation highlights new views and tricks available in the latest versions of JMP. We feature several popular industry graph formats that you may not have known could be easily built within JMP. Views such as super plots, overlay encoding, page grids, doughnut plots, and more are included so you can breathe new life into your graphs and reports!
We will get started. Welcome, everybody, to Pictures from the Gallery 9, More Advanced Graph Building Views. My name is Scott Wise. I've been with JMP for 16 years, and I've had a chance to work a lot around graphing and continue this series. We're going to show you six really neat challenges that we were able to do with Graph Builder, just to expand what you can do with Graph Builder.
Before we begin, we always like to start off as something that will make you think a little bit. I really believe Graph Building helps you make connections. The inspiration for this is there was a British historical educational show that was created by James Burke, and he called it Connections. It went through history and showed how unseemingly, unconnected things that happened really connected with each other and drove big changes. I have a connection that I want to share with you, and I'm doing this through graphing. I got looking at where bats were located in Texas. I was looking at this data, and it led me to triangulation, a certain method for finding distances. That led me to stained glass.
Lastly, I realized that it could help protect landmarks like Notre Dame. I'm going to show you what these connections look like with our data. Here's our data. I had just pulled together in the JMP some areas where we have bat populations within Texas. They could be under a bridge. They could be in a cave. We have quite a few of them here. You might notice I have not only the names of the areas, but their location, what their population is, and even within JMP, just to remind people, you can actually bring in pictures. All you have to do is set the column up as an expression data type. You can bring in pictures and use those for labels or hover elements. It's kind of fun.
I have this information and I went ahead and created a map. Here's a map of all these locations within the central part of Texas, where we have some recording information on bat population. Huge ones, like 15 million down near San Antonio and the Bracken Cave Preserve. Where I live, it's pretty close to this one here. This is the Congress Avenue Bridge in Austin, and we only have a million five there. But this gives me an idea where there are populations, but it's spotty. I would like to get a graph that helped me see more information. I heard about a possible additional graph that's available both on the Graph menu and in Graph Builder. It's called a Contour Plot.
I'm going to go and start over with this information. Let's just go ahead and open up a new one here. Graph, Graph Builder. I'm going to put the latitude on the Y and the longitude on the X. This time, instead of a point, I'm going to keep the points, but I'm going to add this contour. I'm going to change this smoother line to the contour element. You can also see it right up in this area if you want to pick it from the menu bar. I'm going to put the bat population on the color because when you do it on the color, you end up with boundary lines and fill areas.
If I look down here on this contour area right down here below, I can even select the number of levels. I'm going to put that up to 10, and I'm going to drop the fill, but I'm going to color in the lines. This helps my eyes and does a much better job allowing me to look throughout the area and find where there are light densities of that population. But I got really interested in this. I wanted to know, how did they come up with these boundaries? A good hint is if you look at this little alpha slider down here, if I pull it all the way to the right, you can see as I start to move it left, you see it's connecting the points with triangles. It's using this triangulation to actually do this view. That led me to learn a little bit more about it. I'm going to include in this journal, and I will share this journal with you. There's a great blog on our community from Dan Schikore on contour plots within the Graph Builder, and he got to a point where he mentioned, "Oh, it uses this Delaunay triangulation method".
It's a geometry-based method for finding distances, and it was really cool. I wanted to capture these triangles, but it put in here that that wasn't a Graph Builder option.
But we're not out of luck. Turns out that if I go to the main Graph menu, instead of just going to Graph Builder. If I go to the contour plot, they call this our legacy platforms, there's a whole dialog that comes up. If I fill this one in, I'm going to put longitude and latitude on the X and the bat population on the Y. Bat, it now gives me under my red triangle many more options. I can as well specify the number of levels. I'm going to go back to 10. But this time as another option, I can save, and I can save the triangles. Save triangularization. You can see it lists the coordinates for the different positions of the triangle that it used to generate that graph. I thought that was cool, and I really wanted to see if I could do something more with this. When I saved this triangularization, I went back to Graph Builder, and I just put you some points and lines, and I came up with this.
Now, I'm going to turn off the line fill, and you can see what's going on. If I click on this blue point over here on the right, it was enough information to get two sides of the triangle, but it didn't close it.
I'm only seeing two sides of each triangle. So I needed to help this out a little bit because if I wanted to go later and fill it, it wasn't going to work correctly. What I realized I needed to do was actually add that third line, that third side of the triangle. This might be the one side, this might be two sides, but I needed to return it back to the original point. You see, I copy what was the first triangle position just below it, so it has place to go. Now, when I look at my Graph Builder with Line and Points, you can see it looks a whole lot better. I'll turn on, I'll show the control panel, so you can see my settings. Row order does matter, so you want to make sure that is selected. In the fill I'm using, this would be the fill none. You can see it got all the triangles completed, gave them a different line color, which was nice.
On the fill, if you use this fill below, it works out quite nice, and now I have a different color within each of those triangles.
I looked at this, and it looks to me like a stained glass, doesn't it? Like someone used mosaic art or doing stained glass window, and they just use triangle shapes. I started to see there was a pattern here. I went back to my saved out triangulations, and I edited it. I dropped some of the ones that were not significant. This is what it looks like. Now, can you start to see the shape? Why would this help prevent what happened in 2019 in Notre Dame, which was they had a big fire. They just been slowly rebuilding it, I think now all back and open. I figure as much that instead of just using the regular historic stained glass, they could probably find some place to put my nice bat stained glass window. It looks to me like the bat symbol for Batman. If we shine a light through it, if it catches on fire, someone's stealing something, then we can alert Batman, and Batman can come out and save us.
Just some fun with this. But this really was a fun way of expanding more upon what contours can do and see these relationships and even get to the point where it presents beautiful art.
That was just a little warm up for us, but hopefully you saw that contour plots were pretty cool. Let's, without further ado, show you the pictures from the gallery. For the next 20 minutes, I'm going to go through these six views that we picked for this time. This is our ninth edition. Every time our customers challenge us with, "Hey, how do you do this in Graph Builder?"
We're going to look at SuperPlots. We're going to look at Swim Lanes, Overlay Encoding, and Page Grid, our new and JMP 18, Survival at Risk Plots, and then Doughnut Plots. Let's go on to the first one. As I mentioned, I'm going to give you this journal. It will include a picture of what we're trying to do, why it's good and tips to make it. If you want to rebuild the picture, I have the step-by-step instructions as well as I put in your journal right there, the raw data with the final picture scripted in. You can follow these instructions and do it yourself. I'm going to click on this super plot. Super plots are a way of showing a lot of information.
If I take a look at what I have here, I have two treatments in three groups. I have a lot of grouping. That's the data, that's the Y. You can see here, I've even got a little formula here that was calculating in the mean for groups. This 2.8 is the mean for Control A. 5.7 here is for Control B and so forth. I've got a lot of information. Can I get all this on one chart? I'm going to just open up my Graph Builder. I'm going to put both continuous outputs, the data and the mean on the Y. I'm going to put treatment on the X. I'm going to put group on the overlay. Looks busy. It's going to get busier. Instead of just having one set of points, I'm going to right-click here in the chart. I'm going to add a second points element. It's okay because on this first point, I'm going to come in here and under the little variable box that shows you what that graph element is going against.
I'm going to say you just do data. In fact, I could even just call that data points to make it easier.
Then on this one, this is going to be the mean points. On these mean points, I'm going to drop the regular data. It's just showing the means. You can see it's got these groupings where the means are. I'm going to take this jitter limit and go all the way to left to make it go right into the center. Now that I've got this, let's play with the visuals. Let's play with the marker settings and size a good place to do that is under the red triangle or Graph Builder. I'll call these hotspots. You can look at Legend Settings. On this Legend Setting right up here, here's A. It's got a blue circle right now. I'm going to make that a solid circle. B, let's not make it a circle at all. The marker, let's give that a diamond shape. For C, let's give it a square. A is okay for the means, but for the B, let's give that marker size a outlined diamond and let's get C a outlined square.
Just say Okay. Now this looks okay, but things are really competing with each other.
Another hint that's very popular, I go back to those legend settings is where I had the first couple. Let's not make them fully transparent. Let's cut the transparency down below one to 0.25. That way it'll go. I have to do it for each one, but it'll go into the foreground a little better and be ghosty back there. Now I can see where the mean is for this Group B, and I can see where the individual ones are, where the other solid but now transparent diamonds are. I got two out of the three things I want to see. What else did we want? We wanted to get a grand mean for control and treatment. To do that, I'm going to actually add I had right-click, right here in the graph, I'm going to add a bar. But the bar I'm going to use is going to be a float bar. It's got a line just at the mean. All I have to do with the variables is maybe say, "You know what?
Don't do the overlay group. Just do it based off the raw data because it's already set there in those two treatments, control and treatment".
I might even add here for error interval, standard errors. Now I'm done. Now I have all that information from the different groups and the raw data all on one chart. Here's your super track. Again, if you like this, you always have a copy of it to compare yours too. It's here under super plots. You click it, it opens up. I even put a background picture. It made that a little transparent, so it pops a little bit when you view it.
Let's take a look at the next one. Next one on the list is Swim Lanes. Thanks to Jed Campbell, one of my peers here at JMP, he came up with a really nice blog where he was making this nice wooden ball, and he needed different things, different formulas to calculate the measurements, but also a graphical way to see it. He needed to have some channels, some Swim Lanes that helped him figure out, based on what ring he's at, what the dimensions are. To do this, we'll go ahead, and we'll open up the data. A trick to doing this kind of chart, the thing you want on the Y, you want that to be either nominal or ordinal.
Then we're going to compare that to this inner and outer radius. Those are continuous measures here that you had calculated. I'll go in the Graph, Graph Builder. I'm going to pick up the inner and outer radius and dump them both on X. I'm going to pick up the ring and put that one on the Y. I'm going to right-click and change the points to a line. Down here where I have all these options for the line, I'm going to change the connection, and I'm going to make this a centered step. I want to add some labels as well to this. I'm going to right-click in here, go to Access settings. I'm going to turn on some grids, but you notice the grid lines, it's going to be an example of what's going on here. They're going right through the label. You can change that by going up to the Show Tick Labels and going to a Short Divider. Now I've got that, I'll say Okay.
That looks pretty good. Maybe change the line colors to something that are the same. Now we basically have what Jed needs to build his ball.
You can look in Ring 1 and see what the inner and outer radiuses are. I've got more instructions, optional instructions, where you can actually add in, not only a background picture, but you can add in as well measurement increments next to the numbers and add these grid lines behind the picture. They're useful, and it might be for project management, program management where we want to look at things over time. A lot of different uses for Swim Lanes.
The next one. This is a new one in JMP 18. JMP 18 came out a little earlier in the year. Hopefully everybody is converting over to it. JMP 18 corrected a problem we had with overlay encoded. There were times, different views. We wanted that. You had a hard time doing if you included an overlay. Overlay was greedy in 17 seen and beyond. It basically dictated what your lines and your symbols look like, even if you had something in the space for coloring or styles. This is some ice accumulation on the Great Lakes in the US. I'm going to go down here and pick up this data.
A couple of interesting things here. I've got a winter period that's got more recent time versus older time. They were looking at the effects of ice accumulation as a proxy for global warming. Do we have less ice in the winter on the lakes over the months? Let's see what that looks like. I'm going to go to Graph. We're going to put the ice accumulation on the Y. We are going to put it... We're going to look at it by month, and I'm going to overlay it by year. For all the great lakes here, I'm looking at this data. That's getting me a box plot on here. I'm not too crazy about a box plot, I'm going to change that to points. Maybe I don't want all this data. Maybe the summary statistic I want is going to be the mean. Then maybe I'll right-click in here and I will add a smoother to this. That's the view I want. Every line is a different year.
I can see it going across the different months. That looks pretty cool. Now, I was happy with this, but what happens when I add that winter period to the color?
It just goes crazy. I like the color is changed to the different period, but I don't like what it did to those symbols. Now if you go to the overlay encoding, you right-click, say go to the overlay landing zone, you right-click, you get this encoding option, and it says, do you want the coding to follow your Color, your Style, or None? The style is just weird. It's given all kinds of crazy different shapes for the lines. That's really it wasn't what I wanted. Overlaying coding by color. It's got my line colors right, but it's got these individual year, individual points, different colors. I didn't want that. A good thing you can do for overlaying coding is just say None. Now that I say None, now it's exactly the where I want it. Now, I can clean up this legend, in fact, on the one that I included here in our data. What a nice little background picture of a tug boat, or [inaudible 00:21:28] tug boat.
But I got the different lakes on a local side filter. Now, you can see for Lake Michigan, yes, it does look like the blue before periods had more ice and more recent red periods have less.
Lake Erie, a little less. That's more of a southern lake. Not a surprise, you can look at all of them together. Really cool way of using this overlay encoding to get just the view you want for your user. Overlaying Encoding. We're doing with time. We're doing pretty good.
Our next one is going to be the Survival at Risk Plot. This plot here is very popular in biotech, just like the SuperPlots are popular in biotech. This one, they wanted to look at survival over time, but have a table of numbers below it. Where this comes from is if you go look at the data, there is a platform, and its Analyze, Reliable, and Survival. There's different survival platforms that allow you to get over time what's surviving in your test groups, their standard and test. But it's got all these tables of good information. Here's the number that's surviving at the time. I've got all the information I want. But what I can do, I can right-click here, I can make this into a combined data table, and it has all the information that I need from everything that's like it in my tables down here.
So I'm going to go ahead and bring up that data. I just saved the standard in the test, but I've got the survival column and I have the at-risk column. I should be able to go to my Graph Builder really easily. I should be able to put survival on the Y, time on my X, Turn off. I'll just remove the points. Sorry, it got us smoother there. I can overlay by group, and now I, in fact, have that first graph. Now I want that graph below it. A combo graph is where I add another section to one of the axes. In this case, I'm going to take the group. I'm going to go down to the bottom of the Y. You see that it separates this. It's cool. I'm going to do a couple of things here. Where it says Points[Group] down here. I'm going to put the Jitter Limit all on the same line. I'm going to click under the red hotspot, and I'm going to say Set Shape Column by.
I'm going to set it by the At Risk. Now I've got numbers in there. I can play around here and get...
It's where I have the diagonal line. I can get there and get it set up a little clearer in my data. I could right-click in here and, of course, go to the graph and play with the marker size and make it a little bigger, pull this out a little bigger. This was a good one for the filter. When I went to put this here, say version here in your data, not only did I put a background picture, and I brought in some... On the axes, I brought in some grid lines. But you can see as well, I just put some select times out here because I didn't need to see all the data. Now I have a pretty compelling view, and I can see when the last one survived for test, when the last one survived for the standard, and how the different survival curves are affecting my surviving population.
I think we got time if I'm going to go through these last two a little quick, so I make sure we have enough time for Q&A sections later. But this next one is another JMP 18 one. This is a Page Grid. This is setting up pretty much separate charts, putting them in a grid. But in JMP 17, you couldn't get this view. You might notice that Brazil here goes from 50,000 all the way down to zero. Well, this one only goes from 3,000 down to zero. Usually JMP only cared about the biggest scale, and it would force everybody to follow that scale, which we don't want. But now in JMP you have control over that. Sometimes you might want that, sometimes you don't. In this case, I'm going to bring up this coffee data. It's talking about production, exports, consumption. I'm going to put all that on the Y-axis. I'm going to put the year on the X. I'm going to switch the points out to Lines.
Now, I'm going to take the country. I'm going to put it on the page. I'm also going to right-click on the red hotspot here and do a little local data filter.
For the country, I am going to go ahead and just select a couple of countries like Brazil, Colombia. I'll put my place, I have my second home, the Philippines. We used to be a big coffee producer at one time. Then Vietnam. Now that I have those selected, I can right-click up here and put levels per row. This is how you make the grid. I want two by two. I'll just put two in a row. There we go. Let me show you under this red hotspot for Graph Builder. See there's something that says Link Pages. If I went to where it used to be in JMP 17, it shared X and Y for all the graphs. They had to have the same Y-axis and X-axis. X-axis is fine. That's here. But Y-axis now forces poor Brazil to go to 50 on the top end, Vietnam to 50 on the top end. I can't even see a trend for Philippines. They don't have that volume.
It does impress you with how big Brazil is if I'm comparing across, this gives me the same axis. But now I can right-click in here on the link pages and I can say, "You know what? Just share only the X-axis". Now I can see the trend where Colombia is flat.
Philippines, even though they only top out at 3,000, I can see the consumption is really high. It's a good place to put a startup. You have a lot them popping up. I can see Vietnam is really cranking up even more than Brazil in terms of the slope for increased exports. That's the kind of stuff you could do now in JMP 18 with these Page Grids.
I think we have just enough time to show you the last one. It's a fun one. Might go one or two minutes over, but I think you'll enjoy this. We're not big on pie charts in JMP, but there is a type of pie chart called a ring chart, which is great to break into your basic graphs and dashboards. You can use them as selectors, and it's good if you have just a few categorical levels. Here I've got a pie chart, which is the ring variety next to a bar. If I click on something here, it is going to change what's in the bar. How do we set that up? I got this cool data on doughnuts by the whole different generations. You can pick what generation you're in here.
We're going to go and make some graphs. The first graph I'm going to make is just going to be my ring chart. I'm going to take the percent on the Y. I'm going to put the type on the X. I'm going to change points over to a pie chart. Don't panic under pie style, let's go with the ring. I'm going to label this by percent of values. Simple chart, easy to use. We'll I'll park it right here. Second graph, easy to do. I'm going to look at my Gen by percent. I'm going to ask, instead of points, I'm going to ask for a bar chart. Change to bar. There we go. On this bar chart, I'm going to ask for the sum. I'm going to double-click on my axis here. I'm going to tell JMP the formatting is Percent. Add a few grid lines. Now this chart is done and ready. Now, to combine them, the easiest way is just go Windows, Combine Windows.
You can click on the ones you want. I've got the ring chart right here. I've got the bar chart right here.
In fact, by the ring chart, I can make that the filter. When I say Okay, here I go. Now, if I did this correctly I want to see how many people... Let's see, let's do sprinkles. I want to see who's doing sprinkles here. I click on it, and you can see not only... I got this label come up for the picture to remind myself what the sprinkles look like. A sprinkle doughnut looks like Gen Z or the big sprinkles. Who likes cake doughnuts anymore? You can click on them, Baby Boomers, maybe they like to put it in their coffee. But now, doughnut's cheaper than their coffee, right? We don't do that so much. Everybody likes glazed doughnuts. Nice way to have an interactive dashboard. This could be saved back to your data table. This could be saved as a report interactively in JMP Live. Very cool interactively in HTML5. Many different ways of sharing it. Those are the views. I'm going to, again, leave you behind with this journal.
Just go out to where this recording is parked and you'll see my journal. I will also leave you with where to learn more, whether you want a tutorial, look at blogs, other presentations. If you have new views, please go out to the JMP community to the wishlist and challenge us with your graph and challenges. Maybe they'll end up in the next version of JMP, and they'll get into the next pictures from the gallery.
I thank you so much for joining me, and have a good rest of your discovery.