Pictures from the Gallery 11: Innovative Graph Builder Views (2026-US-30MP-2901)

Thank you very much for joining Scott and I as we talk about Pictures from the Gallery. This is our 11th version of this. I thought I'd start with maybe just a little bit… flip the script and do something a little bit backwards. I have an image of a graph that I found on the internet and I spent 23 years in my previous life working in the diamond industry, and we made diamonds mostly for oil and gas exploration, but for lots of other uses as well. We spent a lot of time on this part of the curves, maybe this part right here. I was thinking, boy, wouldn't it be nice to be able to make a graph of just this portion and maybe make it something a little bit first-principlesy? Of course, the answer is yes. This journal will be available, and so you'll be able to recreate all of this. But if I save the image to the desktop, then there are a bunch of add-ins on the marketplace. Scott is going to mention the marketplace a little bit later as well. But one of them is this Image to Table add-in where you can choose a file to upload.

I've got this JPEG of the carbon phase diagram. I just want to grab this portion right here. I guess I should wait for the graph to show up, but I just want to grab this portion here and then fit a model to it and then build a nicer graph. It's really, really simple for me to do that. It just really involves maybe a couple of steps. The first would be recalibrating the scale. I've got my scale, the X-axis goes from 0-6,000 and my Y-axis goes from 10 to the seventh through 10 to the 11th. I hit Okay. Now, all of my axes are calibrated, my data points are calibrated in space here. But then I just need to select maybe a couple of points on this line, grab that element, and now I have selected all these dark points. So it's a simple matter of maybe working through a couple of things to maybe invert my row selection, and then delete these data points that are not that black portion. Then maybe just some careful selection followed by hitting the delete key on my keyboard to delete all the parts that I don't really care about in this image.

We can sneak up on just the portions that I want to have. Then, once I've done that, I have just that curve. Now, the scale is all off and everything, but I can use some JMP platforms to do this. I'm not going to make you watch me do that. I've saved a workflow to this journal so that we can use something like the Fit Curve platform to extract that formula, maybe make a profiler for that, and then, of course, why we're all here is make a graph of just that one portion of it, so I can look at diamond versus graphite and that cutoff point. Point being here is that there's so much available for us when we flip things backwards and grab stuff from images, and then we can use that stuff that we've grabbed and put into graphs. Now, we have put the instructions down here. I'm not going to go through rebuilding this one, but it's just a teaser, maybe something to wrap our heads around a little bit. Scott, you want to take over and introduce us to this year's version Pictures from the Gallery?

Thank you, Jed. Yes. Here's our Pictures from the Gallery. Thank you for that opening case, Jed. Do remember, if you want to see more helpful graphing tools that are available over the marketplace add-ins, we're going to feature a talk on that in the in-person Discovery Summit in Baltimore in October. But this is Pictures from the Gallery. We always are challenged to come up with some views from our users on, "Hey, I would love to be able to do this, but how do I do it in Graph Builder?" Well, here's our six for this year. I'm going to dive in with the first three. We're really excited about them. First one up that we're going to talk about is a radius map. Doing geographic maps are available in JMP. We got latitude and longitude positions, but sometimes it's helpful to draw a radius line, let's say, 50 miles from this center location and see where the other points in the map fall. Now, before to do this, we had to use formulas and scripts. But now, with us being able to pretty much get these radius location points easily through an AI agent, it's really easy to set up if you know how to handle the elements on the graph.

As Jed mentioned, we're going to leave you behind with this journal. It has everything you need from the picture, the tips, the steps, all the way to the data built in along with the finished script attached to the data. I'm going to open up this finished data and show you how we set it up. Jed and I had a bet. The bet was that if we took our home locations, we took Jed's home location in Mapleton, Utah, we wanted to see how many state and national parks we could get to within a drive. What's a drive? About 250 miles. That's about as far as my car will go for it to get charged up or gassed up here. What we did, we just went out, and we asked AI, "Hey, give me all of the coordinates that are 250 miles, straight as the crow flies, from Mapleton, Utah." Those are those latitude and longitude locations. I did the same thing around my hometown in Round Rock, Texas. Here's all that data. It's available for graphing. As you would guess, we came up with all of the national parks and state parks within Utah and Texas.

Now, I gave everything a code so I could keep it straight when I was trying to create this graph. Also, they were nice enough in the AI to give me the measurements of the distance from the center hometown locations. I did come up with a cool little formula, little IF AND formula, just to tell me, is it within the radius or outside the radius of 250 miles? Therefore, I can check to make sure I got this correct on the graph. How will we do this? Pretty easily. We're going to go to graph, Graph Builder. Now, I always get confused on which latitude go on the Y or the X. You don't have to worry about it. Just select both and dump them in the middle, and JMP figures it out. First tip. All right, now I've got points in a smoother line by default. I'm going to change that smoother line to a line. Does that help? Not really. I'm going to go down to this line's element panel. I'm going to put the ordering under row order. I can see the radius coming out. I take the code. I'm going to put the code in the overlay.

Now, believe it or not, this is all I need to set up the graph. But how am I going to turn off these lines that I don't need and make the radius line look smoother? All I have to do is go to my legend settings, and I can do all this through the legend settings, which is pretty cool. Now, I'm going to build my legend as I go, and I know that my B and D are those radius lines, and they deserve to be on the legend, and they're pretty cool. Then I know that A and C were our home reference locations. I'm just going to right-click and give those a different marker. I know that E and F were my state and national parks in our respective states. I'm building out this legend here. What do I do about the lines? Obviously, E and F have a bunch of lines going on, and I don't need lines. I just want points on those. All I'm going to do is click on E and F, and I'm going to do something. I can't get rid of them per se, but I can take them off the graph from our eyes, and I'm going to do it with the line width, ghost them out.

I'm going to say line width of zero. Now they should be off the graph. I will tell when I click this Okay, there we go. I'm in business. One other thing I could do back on the legend settings, I could take the… On the radius lines of B and D, I don't need those to show up. I could take those two points on the radius lines. In this case, I could just go to the transparency, and I can make that transparency zero. Now it's nice and smooth. I'm ready to go. If I right-click, I go on the graph, we go to Background Map. I can choose between a map service or maybe one that's a built-in. I'll just do this built-in US States one. There we have it. That easy to build in your data that's attached. I actually used one of these analyze tools called Tabulate. It's a little pivot table. They give me some raw statistics, and then I used that to make this map, and I added a few labels. Let's see about the back. I can get to 69 state parks in Texas in purple there, and I can get to eight national parks.

Now, Jed doesn't have as many state parks. He's in a smaller state, 28 state parks, and get to seven national parks. But as he pointed out, there are state parks in my town I can't even get to in one single drive, and they're good ones, like Big Bend and Palo Duro Canyon, while Jed can get to some really awesome ones, and he doesn't have to go far, like Zion National Park and getting out here to Arches National Park. We're just going to call it a draw. All right, so that was our first Pictures from the Gallery. Let's see our second, Histogram Box Plot Combo. It is nice sometimes to be able to align a combination of graphs. In this case, a histogram on top and a box plot below that are aligned back to your axis. This lets us see things like central tendency along with the shape of the histogram for different levels of things. This was a requested graph, and it turns out to make it, we're going to have to create some additional graph areas, and we're going to have to use that secret group Y-axis. Where is that thing? It's on the right-hand side of the graph, and I'll show it to you.

Then for free, I'm going to throw in how you can use pictures for hover labels. If we take a look at this, all of our instructions, we're going to pick up our data. This is the Penguins Alt data. To set up this data it's just pretty much these different species of penguins, a different body mass of them. But we do have two dummy categories that were set up. So I made two columns, character nominal columns, and I just threw A into the one and B into the other. This will help us create areas on the graph. Then I created an expression column. On this expression column, I brought in one thumbnail picture. We can just drag them in, JPEG or whatever format, drag them in. I don't need one for every line, just one for one row of the species. That's pretty cool. I'll show you how we can use those. I even turned on the label on those so it will pop up when I hover over a point or ask for a hover label. All right, so let's go ahead and open up our Graph Builder. This one's going to have our body mass on the X.

We're going to put one dummy variable next to it, creating a two-up situation. Then I'm going to put both dummy variables on the Y, and I'm going to stack them to create a four-up. I'm going to right-click in here and change this point to a histogram. I'm going to right-click to this area here, this bottom right area. I'm going to change that point to a box plot. Now I'm going to go down in this area to make the rest happen. While I do that, I'm going to take species. I'm going to overlay by species. Now, that doesn't really have our view. This is another cool view. Now, here's the group Y-axis, and I'm going to move the species as well to the group Y. This creates those swim lanes, and that's nice. Now, what can I do? Down in this area, I'm going to go where it said box plot. You can see that it's getting a little… I got these colored box plots down here. I don't need them colored. I want them to be better sized. So I'm just going to remove the overlay element from that one. That figures that one out.

Histograms are fine. This points A and B, summary statistic, let's go with the mean. This is going to be a place I can hover and pick up a picture. That's cool. Let's for this point, say, let's go and make that a mean, too. But instead of a point, let's go ahead in this red hotspot and set a shape column and make it the species, and then the name pops up. Very easy to do. So there are things I might do to clean up. I would like to make this whole width a little smaller. So I go into where the label is, I pull across until I see the 45-degree slant line. I'm going to just move it like I want to, maybe right about here. That's how I get that line to look. I want to clean up what's on the axes, and I maybe don't need the legend now. Right-click, you can go under Show and turn off things like the legend. You can turn off like the Y-axis. But I can right-click in here as well, go into things like axis settings, you can take out the labels of things. There are a lot of ways you can do to clean this up.

But what about this one over here, this group Y-axis? I don't need it repeated when I have it over on the left. All you have to do here is right-click, go to Show Tree Structure, ask for the properties. Make sure you've clicked on that last group box, and where the visibility is, you just collapse it. That is how I get rid of that edge. Let me show you what my final picture looks like. This is what my final graph looks like. It's nice and clean. I cleaned up everything I didn't need off the axis titles and levels. Now I can just click or hover. I'm going to click, hover right over this little point. Now I can get the picture of the species, and I have a clean graph to look at. The Histogram Box Plot Combo. The last one I'm going to show in the top of our Pictures from the Gallery levels is the Ordinal Dot Plot. This is good if you want to do a high/low constraint view. It's going to make you get a two-level categorical input over some continuous measure. What we're looking on the graph is actually features of a cave, stalactites and stalagmites.

I can see what their count is based on their length, on what's coming from the ceiling, what's coming from the floor. This was a fun graph, and I'll even show you how you do that background photo wallpaper as well. We got our steps down for you. We'll pull up on the dataset. Pretty straightforward dataset. Just go right into the Graph Builder, put my length on my X, put my cave features on the Y, I might color by them. Now, I'm going to go to Jitter Side Ordinals. This came in JMP 19, and it gives me the place that I can then go into the Y-axis and just move those stalagmites up to the ceiling and add those stalagmites. Actually, it's actually easier just to put one at the bottom, then put the other one at the top. Oh, wait a second. Stalactites come from the ceiling. It's got the order reversed. That's easy. Just right-click on the axis settings and reverse order. Now it's got them right. From here, believe it or not, it's just really about you working with the size and type of graph you want to use. I can right-click in here, and I can mess with marker sizes and make them bigger.

Make that marker size a little bigger, get the graphs like I like. You can get the bin spacing like I like. Now, how do I bring in that picture? First off, if you have yourself a photo, just drag the photo right in. That's all you got to do. Right-click, and there's a place to mess with images. So here's the image, I can go to Transparency, I can make that shaded, like a more transparent, like a 0.25. I can right-click again, Image, and then I can go say Size and Scale, Fill Graph. Now I have a pretty cool graph, and you get it just like you like it, and then you're done. So I'm going to turn it back over to Jed to go over our last half of the Pictures from the Gallery.

Excellent. Thanks, Scott. I like those graphs. Something as Scott and I are thinking about these Pictures from the Gallery sessions, there are a couple of different ways that graphs show up in these. One of the ways that they show up is maybe it's a graph that either Scott or I have struggled with to make. So I want to share one of those graphs, and this is a graph that I struggled with. All I was really looking for was a horizontal mean for two different groups with maybe such a confidence interval to overlap them. It was really about showing the concept of power versus sample size. I was on the struggle bus a little bit with this, but it's a pretty easy graph to make, but I want to walk you through the thought process that I was going through. I have… Sorry, I skipped looking at the data. It's really simple data. I've got A and B as two groupings, and I've got my data. I want to get that graph where my sample is across the side. Now, Scott showed earlier that if you add the data points, you can change the row order if it's a line.

That's not available in all platforms, but it's always available. If you right-click, we can actually just create a formula column really quickly by asking for the row. If I have my data and my row, I'm going to drag them to this little automagic zone in the middle, and then I just have to drag this to overlay. I'm mostly done with the graph, except I want those confidence intervals in the straight lines. Now, if I change the lambda all the way to the right, that's still not flat. So that's not what I'm looking for. Maybe more what I'm looking for is some line with a confidence interval. I looked through many of the menu options here to find, is there a way to fit this without a slope, maybe just fitting the mean? I didn't find anything. I thought, well, I know how to actually show a line at the mean, and that's this caption box. If I turn off the lines and turn on the caption box, I can tell JMP to move this caption box down into the graph, but I also don't have any confidence interval spot.

It's funny how easy this is and how close I was with that smoother. I just didn't look hard enough. I was thinking the developers put a lot of effort into adding multiple spline options like penalized splines. I clicked on this moving average one just to see what it was. Strangely enough, I can get that confidence of fit. But if I move that local width all the way to the right, I'm 90% of the way there. So now I just need to drag the trim back. So the lesson for this graph is don't be afraid to click around, to look at options, to fool around with this because you might just find what you're looking for. That's the first type of graph that shows up in this that I wanted to show, and really a lesson being smoothers can do so much. The second type of graph is just a type of graph that maybe Scott or I will see something online, and we'll say, "Oh, I wonder if we could do that in JMP." So this was a graph that I saw where we looked at episodes of a television series versus season versus some rating in the middle.

I like it because you can quickly see maybe some low performers. This doesn't have to be about television, but two categorical variables with some rating. I can also see that as the seasons of The Simpsons episodes progressed, they got progressively worse. They went from green to maybe orange as time went on. So you can see a lot with these. Maybe this would be something where I would want to hover over like Scott showed with those penguins where I could see something, so a first point in a drill-down cycle. But if I look at this data, I stole this from Kaggle, but it's just a list of all of The Simpsons episodes as well as their number in season and their season. So it's very simple, plus there are some IMDb ratings on it. It's a very simple graph to start, and it just needs a couple of little tweaks to make it happen. If we look at our number in season, that'll be our episode number, and we look at our season. I'm mostly there. Maybe I should right-click on this and change my markers to this wide rectangle. Make them a little bit bigger. I fooled around with this and found that eight seems to be about the right size to make that type of a graph with a little bit of space in between.

I am noticing one thing before I get too far, though. I'll put this IMDb rating here in the color. This is backwards from what I'm expecting. It's starting at 1 and going to 34. I want season 1 at the top. Same thing that Scott showed earlier, this reverse order is super nice. You can change that anytime you want. I'm going to change the gradient here to this green, yellow, red. Red is bad, green is good. Only I'm also seeing another problem. When I said red is bad, green is good, this is backwards. So I'm going to reverse those, and green is good, red is bad. Now I'm happy, and this graph is maybe good enough, but I also want the numbers forced into there. That might be a little bit small, but I don't care. It's a bit of a challenge, I want to do it anyway. The secret here, a tip is you can add by right-clicking, you can add multiple elements of the same element. I can add a second set of data points. When I do that, nothing much is going to happen to the graph because I'm showing the same thing twice.

But I can change the first one. This first one was the one that we made already with the almost heat map. I can come here, and I can set that shape column to the IMDb rating, and that will make it so that the numbers show up. Now I just need to go back to the legend settings and maybe make some changes to this second set of data points. First, I'm going to change the marker size to something smaller because I can see that the text is much too large. I'm also going to make the color white. That's going to make them show up on the inside. I'm almost there. I'm just going to change the background color to black to match what I saw on the internet. Now there's just one more little aspect, which is I want the borders and the axes and everything to be black. Scott showed this earlier. When you right-click, and you don't see an option to maybe change a color of something, there's always going to the properties. A little shortcut on a Windows machine up here, in every window is the properties. It's also under Edit properties. Sorry.

I've got this properties panel showing up, and all I need to do is click until I can see depending on where I click, I might see A, the ability to change the background color of something. If I click and see the whole thing, that's even better. I've got this background color, which I can change to black, and I can change the text color to white, and now I am done with this. Super nice. I can close that properties. I have this graph that I saw, maybe hide the legend, and I'm done. That being said, we got one more type of graph that shows up. This is a graph that gets requested a lot, and it really comes down to when people send us data sometimes, they've already pre-summarized it. But something like a box and whiskers type graph is really nice to show data that hasn't been summarized because it can make those summaries. We can see measures of central tendency. It's hard to do when somebody has already summarized the data for me. We're going to go ahead and make a graph with pre-summarized data and also includes something like a box and whiskers type plot.

This is very simple data. I've got years and quarters, they're nested in, so four data points for each year. I've got my minimum sales, my maximum sales, and my median sales. To make a graph, this becomes pretty easy. I'm just going to start by grabbing the year. Then there is a tip here, and that is that the order matters if I'm going to make this type of an interval. It starts by first, we add one edge, in this case, the minimum. Second, we add to this point. We're adding to the graph rather than adding a separate graph. Then my maximum. Minimum and maximum, the key really here is the final one that goes in should be your measure of central tendency or median in this case. If I follow that order, then it's really easy for me to change this to, I know this is a little counterintuitive, to a bar chart, but there are so many different styles of bar charts. One of them is the interval. Now I have these interval plots, maybe TIE fighters, almost like a box and whiskers. Then all I need to do is just overlay by quarter, and I have I'm mostly done with this graph.

The rest is mostly just a little bit of cleanup. So I go to this axis setting. The reason why is if I put in… Well, first, I can't make a distinction in my head or in my mind, I can't really follow where the years are starting and stopping. So I'm going to add a grid here, and the default grid is not at all what I want. I want the grid to be in the middle and between the years. Let me go back to those axis settings. The secret is right here. By default, you show a long mark, but something like a divider is more like what I want. I actually really like these interval trays here because it's a nice visual clue that 1985 is this bucket. Now I have this, and maybe I can go into my legend settings and eliminate some of these middle ones that I don't need and then right-click and move this into the inside left. I have the graph. A really good way to show data that's been pre-summarized for you, and a nice thing, we can see that some of these quarters were not really very well-centered, and others were very well-centered.

Really good aspect and good graph. We've shown you six graphs. What's next? Well, Scott has been nice enough to put together a whole bunch of learning materials with links to the community. There's a wealth of learning right here. If you get stuck, and you're trying to make a graph, these are probably going to answer your question. If not, you can always post a question to JMP's community. If you find something that you can't make a graph of in the Graph Builder, then finally, I think we really want to know, please do submit that to the wish list. Graph Builder is one of the most amazing tools in JMP, and there's a wealth of information and innovation to be had from being able to tangle with the data and view it in lots of unique ways. Anything to add, Scott?

No. Challenge us. Remember, everything you do within JMP can be saved as a script, as a workflow, so you can reproduce it. We have a lot of add-ins that even go beyond what you get within the Graph Builder menu to help you. So challenge us and give us your ideas. We look forward to the next batch, but we're hoping these Pictures from the Gallery are very useful.