Qlik Sense June 2017 is Finally Here!
The new version of Qlik Sense has been presented to the unwashed masses. There are 3 new visualizations we will focus on today, but there are several other great additions to the platform that are worthy of exploration:
- 3 New Visualizations
- Improved Data Preparation
- Script-side dimension color assignment
- On-Demand Apps
- The new Analytic Connector – Will connect third-party engines like R to the front-end Qlik Sense experience
Today we will take a look at the new promised built-in visualizations. They are:
- Distribution Plot
As with all Qlik Sense visualizations, the key to using these new visuals is really in understanding the use-cases for the specific data visualization, rather than having any particular knowledge about using Qlik Sense itself. All 3 of these visualizations have very specific use-cases related to the distribution of the dimensional values.
I have used a previously featured app that analyzes world-wide malaria data.
The distribution plot is great for displaying dimensional data points distributed along a horizontal line by the expression. So, for example, several low values will be bunched at the left while the high-value outliers will be spread out along the right side of the chart. Rather than seeing only a total, the distribution plot charts individual values spread across a plane, like mayo spread across a slice of bread. You can add a secondary dimension to create multiple horizontal planes.
There are a few tweaks that you should be aware of:
I like the idea of having the largest secondary dimension sorted to the top of the chart. To do this, I simply took the expression for the chart and aggregated it by the secondary dimension.
Also, I added a reference line to give us the average for all the regions. This was done in the Add-ons > Reference lines area.
Rather than seeing only a total, the distribution plot charts individual values spread across a plane, like mayo spread across a slice of bread.
The solid color tells us the range of the primary dimension values while the bubbles tell us the individual values of each value. I have not yet experimented with a large number of dimensional values, but this might cause performance and/or usability issues. My initial thought is that the distribution plot is useful when an individual line is limited to 100 or less individual values.
The boxplot becomes interesting for statistical applications. Over a dimension, it calculates the expression and finds the distribution of the values, focusing them generally into quartiles where 25% of the values, for example, fall into an area of the chart. The boxplot comes with automatic options for Standard (Tukey), Percentage, or Standard Deviation plots. The whisker length can also be configured with the default being 1.5 quartiles from either side of the box.
You can definitely use a secondary dimension to create multiple boxplots in the same object. It did not really make the greatest visual with my dataset, but as with the distribution plot, the primary dimension builds the plots, while the secondary dimension splits them into separate sections.
You may choose to remove the whiskers and you may also choose to orient the chart vertically.
The histogram is a way to group the frequency of a numeric dimension into a range. For example, how many values were between 1 and 5, and how many were between 6 and 10. It is a great way to visualize numbers that fall into a normal distribution. My data did not fall into a normal distribution.
The requirement for this chart is simply a single dimension. It must be numeric which means that when you go to add a dimension, you will only find fields in the list that are numeric. In this case, I went ahead and created a field in the script that did the row-level math to determine the cases per 100,000 before using it in my chart. It will count up the rows into bars according to the range that you assign it in Histogram Settings. Because my range went from 0 to 45 thousand, I manually configured the Number of bars to 9.
It seems that these visualizations fall into the theme if “Distribution Charts”. They are worthy additions to the Qlik Sense platform and as a bonus, will be fully supported by Qlik.
I was getting worried since every answer to “Will we get <insert visualization name here>?” seemed to be “Build the extension yourself”. But this is definitely a welcome set of additions to the platform.
What are your thoughts on these visualizations? Were there options you needed that were not included? Were you hoping for a different object?