Skip to main content
Announcements
Have questions about Qlik Connect? Join us live on April 10th, at 11 AM ET: SIGN UP NOW

STT - Visualization Solutions for Qlik Sense

cancel
Showing results for 
Search instead for 
Did you mean: 
Troy_Raney
Digital Support
Digital Support

STT - Visualization Solutions for Qlik Sense

Last Update:

Feb 10, 2023 7:53:56 AM

Updated By:

Troy_Raney

Created date:

Feb 10, 2023 7:34:48 AM

Troy_Raney_0-1676032337691.gif

Environment

  • Qlik Sense Enterprise on Windows
  • Qlik Cloud

 

Transcript

Hello and welcome to the February edition of Techspert Talks. I'm Troy Raney and I'll be your host for today's session. Today's presentation is Visualization Solutions for Qlik Sense with our own Patric Nordström. Patric, could you tell us a little bit about yourself?
Yeah, thanks. Really nice to be here Troy. So, my name is Patric Nordström. I'm the Product Manager for Qlik Sense Visualizations and the Analytics. I've been with Qlik since 2017. I spend half of my time talking to customers and the other half trying to guide the development for the product. I try to tell people how to use all the fantastic features that we have, and how they can be combined to achieve new things.
Great, and I love how you take what you hear from customers, and questions they have, and concerns; and turn the answers into solutions - actually in apps and share them. I hope you can highlight a lot of those today. Now, I understand a pretty common question from customers is how to visualize dividing data into groups. Do you have a solution regarding that?
Yeah. As the product manager, you get a lot of questions. It's a lot about explaining how you could use the product to achieve different things; and my favorite answer is typically the answer to the question: which products or which countries contribute to the first 80 of our turnover.
Right.
The answer could be a Pareto chart which visualizes countries divided into different classes. So, A, B and C have the big customers that represent the first 80%, then you have the next 10, and the rest of the 10%. And it's quite easy to do that in a Sense. You can either look up a blog post by HIC or somebody but could also use the Insight Advisor to create it for you. It's one of the analysis types called the Relative Importance.
Oh.
It is of course, you don't have to use a Combo Chart, you could also use a Pivot Table. In this case, I made it a bit more fancy because I made it possible to pick which should be the basis for my classification. If it should be Country or if it should be Categories or if it should be Products.
Oh, that's really cool being able to switch between them. So, that's a Cyclic Group that you set up?
Yeah. We do have a group concept in Qlik Sense which is the Drill Down Group where you have dimensions in a hierarchy. If you just want to cycle through different dimensions, you could just use the Variable and the Action Button to do that which I've done in this case.
Okay. So, that's an Action Button? It just cycles through Variable?
Yeah. There is also a recipe for how to do a Cycle Group as well.
Yeah. We'll definitely include links to everything that we mentioned here today along with recording of this session. I'll be sure to include the link where people can find this app in the 24 Days of Visualization season 2. All right, well, what's next?
Yeah. Another typical question that I get is: which combination this is the most popular ones it shows up in? So, many different variants. People ask me what kind of connectors are mostly often used together on our Cloud solution, or which combination of our products is the biggest seller? First, find a way to describe the combinations, and I do that with Concatenation. So, I just make a Distinct Concatenation of all the types. In this case, in my test data set, I have different categories per order ID, and then I group them together: Childresn’s Wear, Menswear for instance, then I turned that into a dimension instead by using the Aggr function.
So, I use this expression Concat all the types in a Distinct manner.
Okay.
If I do that, I have the frequency of the combinations instead and then I put that in a table and I count the number of orders for each specific combination. So, in my test data set, the most common combination is that you have only items for one single category.
Okay.
As a bonus, I added an image to show the combinations. Each one of the categories I have a number and also a color associated. It’s a dynamic SVD; not that hard to make it actually. Just one box of colors like that.
I think that's great with the image, because it instantly provides some insight. Now I know another common ask is for a chart to display all values (like in a calendar) even when the measure result is null. How would you get around that?
Well, there are several different types of null values.
Yeah.
I won't dive into the specifics. In this scenario, we have Sales. But for one specific country, we don't have any Sales during March, April, September and December.
Exactly.
The normal way that most charts in Sense works is that it omits the ones that doesn't have any values associated to it. You could use Set Analysis expression to force a 0 bar for those dimension values. And what this means is that it will ignore the selections. It requires that you have some sort of Sales for any country during these years, and then you add Sum of 0. So, it will be turned into zero for those specific values.
So, then it's basically it's 0 instead of null; and so, it actually will show something?
Yeah. For those of you who are familiar with QlikView, this is how it's done behind the scenes when you click that tick box called Show All Values; it actually adds this. But Show All Values is quite common to use, especially if you have several Bar Charts at the same time; typically, if you have like a trellising or mini-charts that you have Bar Chart for several countries at the same time, and you want to compare them. Then you most cases want to make sure that you have the same dimension values and the same axis scale in order to be able to compare them.
That's great. I was hoping we could jump to your solution for adding Annotations inside a chart.
Yeah. As a developer, you want to provide some kind of text inside of your chart. The way in you can do that today is to use dimension-based reference lines.
Okay.
We have already measure based reference lines that you can put on the y-axis, but you can also put them on the x-axis; which is interesting to do if you have a time axis. As a typical example, that you want to highlight what happened before an increase of Sales or something that you had a campaign you want to explain; you want to add something to it. Of course, I tried to make the text relevant even if you change the selection or the data. So, you choose the things that you want to highlight by doing that. One thing that you can add for an annotation is to indicate if you have Partial Data. And that could be quite useful as well.
That's super valuable. I had some presentations just yesterday where I could have used something like that.
Yeah. Yeah, I think we spoke about this last year as well, to make the text that you add to your or dashboard relevant. Any text that you put in you can make it Dynamic.
So, it will respond to changes in the selection.
That's fantastic. I know a pretty common problem that customers have is they want to be able to share it with their colleagues; and being able to export it to Excel or other platforms is a pretty common problem.
Yeah. That's super common that. Normally, when you use the dashboard, you're on your big screens. And it works fine, but as soon as you want to perhaps share it, you want to add it to a PowerPoint presentation.
Exactly.
You have a different audience, people are sitting in the back, maybe they want to be able to read what's on your chart. And a super easy way to do that is to make sure that you hit the browser zoom before you make a screenshot of that chart. The difference between 100% zoom and 150%, it's quite obvious that this chart to the left is much more easier to read than the one on the right.
Yeah.
But it's so common; but when it comes to the images and charts, they tend to forget that you have to make everything very large to make it obvious.
Makes so much sense when you say it, but a lot of people don't think about it.
So, it's really good that we highlight this. And it's critical really, because there's a big difference between doing your own analytics on your own large screen versus sharing it in a zoom meeting or in a presentation where sometimes what people actually viewing it on might be a smaller window.
Yeah. Exactly.
Now, while we're adding new things to the product all the time, but I know that Pivot Tables are still very popular. Do you have any tips with getting the most out of this kind of basic chart type?
Yeah. So, the normal way a Pivot Table works is that you set up a sorting for the different categories. I have my countries in alphabetical order, and then I have my categories in a certain order.
Right.
They are the same, independent on which country. Choose Ireland, Baby Wears first; or I chose Canada, you still have the same order. aAnd that is normally how you would like the Pivot Table to work. But if you want to sort by the Measure instead of the alphabetical order of the dimensions in the Pivot Table.
Right.
And there is a way to override the default setting of the Pivot Table which has a global sorting of each level in the hierarchy, and it's called Sort by First Measure.
Can you show us where to find those settings?
So, under Sorting there's this tick box called sort by first measure if you have several measures you have to put the one that you want to sort on as the first measure and then it will make the Sorting of the individual categories followed
Okay. in this case, it's sales
Yeah. exactly and it's numeric .
So, the order of the Categories can be different depending on how big the Sales value is. In this case, I highlighted also that I'm actually doing Sort by Sales by adding a Unicode Arrow to show the sort.
Oh, that's smart.
We are currently developing a new version of a Pivot Table that will be probably released later this year that would have more abilities to do interactive sort directly in the view mode. Now you have to be as a developer to go in to decide how the Sorting should be done.
So, currently users can’t double-click to reverse the sort order or anything like that?
Yeah. You can't do that today in the Pivot Table, but the new version will have more interaction features. Sometimes you want the user to be able to manipulate the Pivot Table and sometimes you don't.
Yeah.
So, in this case, it's a matter of the app developer to decide how the Sorting should be.
Next, could we talk about setting colors in charts?
Yeah. The default behavior if you do color by measure is to have this type of gradient or classes and it will adopt to the current selection. So, if I make a selection, I will still use all the colors. So, the colors will be different.
Hm-hm, very dynamic.
Yeah. And sometimes you don't want that. Perhaps you want to have a fixed color scheme. In this case, I'm looking at elevation contours, and I want to have the same type of elevation the color for each, even if I make selections.
Right.
And there's two ways to do that.
Okay.
You could set up a Master Measure. The Master Measure in this case, is elevation; and there's something called Segment Colors.
Ah.
And then I set up limits for each one of the class breaks that I want .
So, then you could reuse that Master Measure.
I'm doing that in the Map Chart here.
Okay.
Another way to do it is to create the Dimension out of the Measure, and then assign the colors to the values. So, in this case, I use the Master Dimension, and in this case, I'm using a Master Measure.
Okay. Well, that's really great. I feel like this has come up before in conversations that I had, people asking how they can assign specific colors instead of having a dynamic. Because dynamic’s great, but not for every use case.
Yeah. I'm a big fan of using fixed colors, because you definitely confuse the user if you start fiddlering around; that one color means something else if you just make a selection or if you move to another sheet.
Yeah.
If you decide to use colors, you should use them wisely, because it's easy to get lost and just add to confusion rather than to try to be more clear or precise. It kind of contradicts what you aim to do in your presentation.
Exactly, and I know some customers have asked to have specific background with their design of their apps.
Yeah.
Have you found a solution for that?
So, that's one of the new features that we rolled out last December is the ability to set up a sheet background. Under Styling, there's now a Styling section also for the sheet where you can set up a specific background. You can also pick a background image from the media library, and you can decide on how to place it. You have to ask yourself a lot of questions: whether does it really add value to my dashboard? Does it make it clearer? Perhaps that's why it took a while before we added the possibility to add an image to the dashboard, because in many cases it only adds to the clutter. But I'm the first one to admit that a nice background can really lift the dashboard and make it really nice to work with; and you want to have higher adoption, that people should like what you do. And if that's the effect of having a nice background, I'm all for to use background.
And this is a sheet specific setting. Right? So, you could have the background on the cover or some specific sheets where it's more…
Yeah.
Useful and not on everything?
Yeah. You could use the same for the whole app if you want, or you could have different themes on different sheets. Kind of ties back also to another feature that we released: the Fine Grid settings. We used to have quite few settings for the grid settings. We had Wide, Medium and Narrow. Now there is also a Custom setting to position objects more freely. So, the grid is quite a common design tool. Of course, you should think how you use the structure and how you place things. Maybe not shift between so many different designs in one single app; because that also makes the user confused. Of course, it simplifies to be consistent. But sometimes, you have a very specific need that you want to do something specific, and this new setting allows you to do that. It's connected also to reporting. The sheet background and the Fine granular Grid also allows you to do printing on the background as well.
Great. Well, before we run out of time, I was hoping you could highlight a couple of chart solutions, starting with the Slope Chart.
Yeah. In Qlik Sense, there's this set of standard charts, but there's also many variants of a single category of the charts. So, the Line Chart for instance. Now, I'm plotting Categories by Sales per Year. If you want to turn that into a Slope Chart, the idea with the Slope Chart is to show the progression over two specific time points.
Okay.
It's good for showing trends. Instead of having all the years at the same time, I highlighted two specific years. So, from 2019 to 2022.
Right.
I can see that Sportswear has dropped from 25% to 15%. Good on highlighting the more drastic changes in my data. And I'm only saving the maximum and minimum of the dimension values. And in order to make it a bit clearer, I use the Value Labels instead of having a y-axis.
Right. It tells a very clear picture. The one on the left, it's a little bit muddled, and this is like trying to express a specific point, tell a story and I can see the effectiveness of it.
Yeah. In many cases you have to remove things in order to highlight others.
Exactly.
Perhaps that's why we have so many different charts to pick from. They all have a specific purpose to highlight a specific feature of your data.
Now if we have a little bit of time, I would love to be able to squeeze in how you can create a Pie Chart with a hierarchy. And I I love the idea of the Sunburst Chart that you had there.
Yeah. Sunburst Chart is a typical example of what happens if you have too many categories.
Yeah.
It's basically just two pie charts.
Okay.
They are placed on top of each other. And in this case, I have my Product Categories and I have my Product Names, and they are linked together. I’m coloring by category. So, here's all the Women's Wear, and then I have the individual Products underneath. The Women's Wear, of course it becomes easier to read if you don't have that many categories to look at at the same time. This type of hierarchical charts like the Sunburst Chart or the Org Chart or the Pivot Table; kind of works nicely together. They are good for showing the overview. But there's a sweet spot of Dimensions that you could use; perhaps somewhere between 20 & 50 Dimension Values is probably the maximum of what you can place into a Sunburst Chart.
Yeah.
This specific Sunburst Chart I used as an exercise to create a chart from scratch. It's based on the Map Chart. The map Chart is basically just an advanced Scatter chart that you can place different layers with lines, points and areas on top of each other. Then I just lay them out that way. To build a Pie Chart, you need to figure out the Accumulated Sum of the different categories, and then I use RangeSum and Above to get the Start and the End percentage for each one of the categories; and then I use that to place the label for each one of the categories.
Fantastic.
But it's quite nice, especially if you limit the number of Dimensions.
Yeah, exactly; keeping the number of Categories down really makes this valuable.
Yeah. I have another tip on how to use the Org Chart for similar analysis.
Okay.
The Org Chart is tailored to show a company structure. You have who's reporting to who; and then you place it into this tree, but you can use it for other purposes as well. If I have Sales, and I want to see how they are spread out in different Product Categories; and they how they are spread out to the different Products; the Org Chart has a nice interactive mode that I can expand and collapse, different nodes, I can zoom in and I can zoom out.
Yeah.
The Sunburst Chart really wants to be fully expanded all the time, but the Org Chart has this feature that you can expand and collapse it; especially if you use it for Tasks and Subtasks. This is just a test data set with different tasks when constructing a house. I don't have to have everything expanded. I can go into the different phases, and I can see the Completion Rate, and I can investigate it further.
And the Org Chart, is that one that's out of the box, or is this something you developed?
Yeah. The Org Chart is in the Visualization Bundle.
Okay.
It has a table of nodes; and the nodes should be referencing themselves.
Okay.
It's quite useful; you can reuse it for many different purposes.
Now it's time for Q&A. Please submit your questions to the Q&A panel on the left side of your On24 console. I guess I'll just take them from the top. First question: is it possible to work around the current limitation of points displayed in a line chart? 3000 dots I believe was the question.
Yeah. As you may have seen, we have increased the threshold for the Scatter Plot.
So, you can tweak it, because it falls over to a Heat Map presentation if there are too many dots. And we're working on something like the same for the Line Charts, that you can push that threshold up when you are displaying things in a Line Charts. Actually the data page that gets sent to the browser increases with every measure that you add. So, even if you only have 3000 dots, you might have several measures, and you might have a Secondary Dimension, and you might have a Tooltip or something. So, it quickly becomes a lot of more data points than you expect. But we are working on something.
Okay.
Meanwhile of course, it's very hard to find a human who can distinguish between 10 000 points in a Line Chart. It is of course always better to filter the data. That's what we do with data visualization, that we always try to throttle the information that we sent to the user. Basically, you provide an overview first, and then details later using drill down, do using zoom in or a tooltip or something; instead of brute force, sending a lot of data points. It's always good to consider: is this viable for my end-user? Can he really distinguish between the different values?
Yeah. That's a good point, because they all need to provide insight, and if it's just overload then…
Yeah. Yeah.
It's not effective. Okay. Next question: would you kindly highlight any newly available chart objects or visualizations?
Yeah. As you may have seen during the last milestones, we have been very occupied by adding a lot of new functionality to the Charts. Right now, we're in a flow of adding settings for the titles, the subtitles, the footnotes, the axis labels, the value labels, the background of the chart, and the color background of the - it's a lot of small settings that you can combine the latest that. We added functionality to the Scatter Plot, which we added the best fit line.
All right.
Also, I want to highlight my Top 10 Tips apps, because many of the charts that you can derive; such as an Icicle Chart, or an Area Chart can be constructed using the settings inside of the standard charts. On a blog post that I have called ”Let's Make New Charts with Qlik Sense,” I posted recipes for every one of the new charts: how to do a Chord Chart, Coxcomb Chart, Activity Gorge, Violin Plot, Stacked Venn Chart, all of these are possible to combine in the existing settings to add to Qlik.
Sounds great.
29 standard charts and now it's 73 we come up with.
That's fantastic.
Yep.
Great. Okay. Next question: quite some customers want filters directly on the dashboard how to do that neatly, so it doesn't take up so much space?
So, my best tip for that one is to use a Filter Pane, but you make it just 1 grid box high, then you modify the title of the chart to use the Concat Distinct and a pipe between it. I mean these types of horizontal filter panes works well if you have a limited number of Dimension Values, like 4 Quarters for instance; or Years maybe, you have just 2 different Years or maybe Months. So, they work nicely for that. If you have 1000s of different values of course the horizontal presentation is not that good. We are this Spring launching a new filter pane that will have a horizontal view directly. So, you can display the values in a grid mode or a horizontal mode. That would probably be one of the new features of the filter pane.
Great tip. Moving along, next question: can you widen the Master Item objects column?
So, you can't; but see if you hover across them, you will actually see the full name of the master item. also if you search for it you can also limit the amount of master items. So, a combination of Hover and Search gives approximately the same functionality.
Right.
But I I can agree that we should have resizable panels more often. We are doing changes to the client. So, it might be that we can fix that in the future.
That's a good workaround. Next question: do you have a suggestion for how to display Regression Analysis? Is this possible with Qlik Sense?
Yeah, Regression Analysis, you can either use the formulas for regressional analysis and compute it yourself, and then you can present it any way you like, in a table or something. Maybe you are referring to the Line Chart option to be able to have a Trend Line; where you can add a Trend Line, and you can decide the - perhaps a 2nd degree. The regression analysis, it actually tells you when you hover over the line. There it is.
Yeah.
Okay. Someone had a question about: how to add a smaller text to a label, like a sub or a superscript?
The settings for the data labels or the access labels is now in a custom Theme.
So, if you want to have smaller text in an app, you write a new theme. It will actually change the data labels for you. So, that's one way to be able to change the font size
Okay.
We are adding settings to the Charts. So, it will be possible in the future now also to going directly into the Styling Section and under the chart; als, change the title, the color, and the size of the axis labels; and access titles.
That's great.
Another way to add the things to data label is of course to add it in a Custom Tooltip.
So, when the user is hovering over a value, he will get additional information.
That's a great tip. Okay. Moving along: if the user's browser zoom is too high, some visualizations don't appear correctly. Is there a way to have a message pop-up instructing the user to adjust their zoom if it is too high?
So, the browse zoom is something part of the browser. So, it's not possible to detect whether the customer is using a browser zoom or not. All the charts are constructed in a way…
Yeah. They are dynamic
Responsive so they will actually change. So, it's hard to define what is wrong also. If you want to have a fixed size of your chart, you can go in and change the sheet size to instead of being Responsive to be Custom.
Okay.
Then it will always look the same, even if the user is zooming.
Nice to know how to affect that behavior.
Yep.
Moving along. I see a couple more questions: would you show a good way to adjust the color of a KPI based on the measure?
Yeah. In most cases, they are talking about having text in a KPI box that they would change on some kind of status.
Right.
So, the the conditional coloring of a KPI, if you go in here, and under Color, Conditional Colors; then you add Limits, just as you do with Conditional Coloring in Master Measures. You can add that to your KPI, and if you want to use this functionality, the value that you put into your KPI box has to be a numeric one. Otherwise, you can't set up limits like this. So, now I'll set up limits for 2 and 3 so they will be become green; above 3 yellow; between 2 and 3; and red if you are below 2.
So, one way to do it is to use ”Dual” to make sure that for my status new complete processing. I have also, a code that means something. So, if I set it to complete, it will be number 3. If I'm still in processing, it will be 2, and then I can have the coloring to be dependent on values so to speak.
That's great.
And the easiest way to do that; you can also set that up in the Master Measure section. Either you decided it to be in percentage or a fixed limit. So, you can set your up your own color scale for things, and then you can reuse it in in several charts; which is of course very good for the end user if the colors mean the same, especially on one chart, across many charts, but also ideally the color would be the same of the whole app. Otherwise, the cognitive load on the user when he has to interpret every chart. So, when he comes to a new chart the coloring means something completely different.
Yeah. All right, moving on to the next question: is there a way to limit a map chart to show only data points inside a set parameter to keep the map small? I think they mean if even if they have data points outside the existing map, to limit the map to a certain area.
So, one way is to use something called Clustering so you group things together. So I have for Europé, I have 1252 airports; and then I select that cluster point; and then I get another layer of cluster points; and then I can select those.
Okay.
In the end, I will have the individual airports in that case.
So, is that a dimension that you add, the cluster?
Yeah. A cluster in this case is another layer of information.
Okay.
This is some kind of summary for a lot of data points at the same time. This is one way to do clustering. You can also create some kind of hierarchy, that you first select the Country and then you display the underlying information.
Okay.
But in this case, I'm just clustering the data points.
Great solution to that problem.
It's called ”Kmeans;” K means clustering.
Okay.
It's an engine function.
Great. And we'll have a link to this app as well if people want to dig a little deeper and see this example. Okay. Last question: could you show us how you built the Bullseye Chart in your app? I guess somebody saw it on that title sheet.
Yeah. The Bullseye Chart is basically a hierarchical chart that shows (in this case) projects in different stages; which is called live study and trial, which is the colors of the projects; and then they have some kind of segmentation going on, that they are divided into different categories, and then they also have this time function. In this case, so they have 4 different time scales where they group the data together. I'm using the Map Chart in this case, with none as base map, and then I added a line layer for the arcs and are using a helper dimension for the line breakpoints to get the segmentation going on and then I just added Point layers for the year labels and these project bullets.
Okay.
So, the center of the bullseye is 0 x 0, the origo of that one, and the outer rings stretch from minus -4 to 4. So, bullets are lined up along the arcs counterclockwise with an offset from the category start. Of course, this only works if the number of dimensions values is somewhat reasonable. In many cases, for this type of infographic charts, they have a smaller window of usability that they can work well, between maybe 5 to 40 projects otherwise it becomes way too cluttered. But it's a nice way to organize information, or a new way to show hierarchical information in some sense.
Very cool. Well, Patric, this has been really great. I always love getting to hear from you; and to show off all of these solutions to our customers visualization challenges. Thanks so much for your time.
Thanks a lot Troy. It's always nice to come here. Hopefully it gives some ideas on how to use the product. Qlik Sense is so flexible a tool, and you can come up with so many different possible solutions to the same problem. So, I hope did my part to spread the word on how to use the product. Thanks a lot.
Okay, great. Thank you everyone! We hope you enjoyed this session; and thank you to Patric for presenting. We always appreciate getting experts like Patric to share with us. Here is our legal disclaimer; and thank you once again. Have a great rest of your day.

Contributors
Version history
Last update:
‎2023-02-10 07:53 AM
Updated by: