Press "Enter" to skip to content

Category: Visualization

Conditional Formatting Line and Area Charts with Power BI

Soheil Bakkshi shows how we can conditionally format line and area charts with Power BI:

One of my customers asked me to show time series in line charts and area charts. But she want’s it to be conditionally formatted based on the average value over time. Let’s keep it simple, she wants to show “Sales by Year Month” in line chart, but, highlight the data points that are below “Average Sales per Year Month”. As you may know, we currently do not have the luxury of formatting line charts and area charts. But wait, this post is all about that. Let’s dig into it.

From the above scenario, you perhaps already guessed that we need to create a measure which defines the colour based on “Average Sales per Year Month” to be able to format the chart conditionally. If any data point is below the “Average Sales per Year Month” then we highlight it in Orange, if it is above the “Average Sales per Year Month” then we stick to the default colour.

Let’s do it.

This is definitely not straightforward, but once you see the process, it’s pretty neat.

Comments closed

Designing a Layered Donut Chart in Power BI

Prathy Kamasani shares how to build a layered donut chart in Power BI:

You can view the full report here – https://prathy.com/portfolio/school-uk/

The tricky bit of any good data visualisation is finding interesting data, inspiration and story. In my report, my story was to show the comparison of Ofsted school ratings among the total number of schools. I also wanted to carry my story between report pages with colours.

So in this particular visual, I wanted to make it visually appealing, so I decided to show two metrics on each donut visual, the measure I want and the total number of schools. Then I resized each visual in a way so that they look like one visual. Those days there were no grouping, but now we can group them as one visual too. Also, when I first designed this report, default Power BI donut chart didn’t have an option to resize the ring, so I went for Circular gauge by MAQ software. Now we can use the default Donut chart too.

Click through for more details.

Comments closed

Counting Tidyverse Package Arguments

Theo Roe has fun figuring out which tidyverse packages have the greatest number of available arguments in functions:

Before we start anything, I’d like to mention that most of the hard work came from nsaunders and his great blog post Idle thoughts lead to R internals: how to count function arguments.

Let’s get started.

The aim of this blog is to capture the number of arguments present in each function with packages of the tidyverse

Click through to see the code, as well as some methods of visualizing the results (methods which you can use in other situations).

Comments closed

Custom Power BI Visuals with Charticulator

Kasper de Jonge shows how we can use Charticulator to build out custom visuals in Power BI:

Recently I attended an excellent session by Andy Kirk on the state of data visualization for 2019. One of the tools Andy is most excited about is called Charticulator. For those of you that haven’t heard about it, Charticulator is an open source project from Microsoft Research. Using a web UI you can design almost any charts by interactively specifying constraints (NO code).

The best news is that you can use these visuals directly in Power BI. You can even use your own data coming from the PBI data model. This came as news for many at Andy Kirk’s session so decided to do a quick post on it. Building very custom charts entices a lot of data artists but it is often hard to bring this to main stream BI product. The ease of the Charticulator and Power BI integration brings this type of data visualizations to a much broader audience.

This is a lot easier than telling people to learn D3, though Charticulator will necessarily have more limitations than writing all of the code yourself.

Comments closed

Realistic-Looking Islands with R

Holger K. von Jouanne-Diedrich uses fractal math to create realistic-looking artificial islands:

Here we will turn this principle on its head and use it to actually create realistic-looking landmasses with R. The inspiration for this came from chapter 4 “Infinite Detail” of the book “Math Bytes” by my colleague Professor T. Chartier from Davidson College in North Carolina.

The idea is to start with some very simple form, like a square, and add more detail step-by-step. Concretely, we go through every midpoint of our ever more complex polygon and shift it by a random amount. Because the polygon will be getting more and more intricate we have to adjust the absolute amount by which we shift the respective midpoints. 

Click through for the code.

Comments closed

Using PowerPoint to Create Power BI Layouts

Jon Fletcher has a good tip for snazzing up a Power BI dashboard:

First question, why bother with layouts?
Using layouts in Power BI allows a user to make their visuals stand out better, the page looks professional and more appealing to its audience.

Second question, why PowerPoint?
The default page size in Power BI desktop is 16:9, (this trick doesn’t work for other Power BI page sizes), which is identical to a PowerPoint slide.
Therefore whatever is designed in PowerPoint will fit onto a Power BI page perfectly. Also PowerPoint is very easy to use; most people are familiar with it.

Click through for an example. It’s easy to go overboard with this, but Jon does a good job of using a muted color so that the edges don’t overwhelm your eyes. I might knock it down a shade or two further from that, but regardless, this is a nice tip.

Comments closed

United States Maps in R

Laura Ellis shows how to use the usmap package in R:

Today, I’d like to share the package ‘usmap’ which enables incredibly easy and fast creation of US maps in R.

In honor of US Thanksgiving tomorrow, I’m going to make this blog Thanksgiving themed! In this tutorial, we will use the gTrendsR package to pull US Google search results on the keyword “thanksgiving” and plot the popularity by state.

Click through for that demo, as well as links to more demos on map usage.

Comments closed

Visualizing Kafka Data Using D3

Mihalis Tsoukalos extracts, explores, and visualizes data (with D3) from a Kafka topic:

Now that you have your data in JSON format, you will use D3.js in order to visualize it. As JavaScript code is embedded in HTML files, the final version of the D3.js code can be found in visualize-spatial.html, which contains the following code:

D3 is extremely powerful, though that power comes with a fairly steep learning curve.

Comments closed

Fun With Waffle Plots

Sebastian Sauer has a two-parter on waffle plots. The first part is an introduction:

A waffle diagram is a variant of (stacked) bar plots or pie plots. They do not have great perceptual properties, I’d suspect, but for some purposes they may be adequate. This is best explored by example. This post draws heavily from the introduction of hrbrmstr to his Waffle package.

The second part uses emojifont to show pictograms as well:

A Pictogram may be defined as a (statistical) diagram using icons or similar “iconic” graphics to illstrate stuff. The waffle plot (see this post) is a nice object where to combine waffle and pictorgrams. Originally, this post was inspired by HRBRMSTR waffle package, see this post, but I could not get it running.

Maybe the easiest way is to work through an example (spoiler: see below for what we’re heading at).

This type of plot doesn’t work for everything, but I can think of a few places where it’d be the right choice.

Comments closed

Decomposition Trees in Power BI

Tomaz Kastrun takes us through a new visual in Power BI:

Decomposition tree is a data presentation of slicing and dicing of selected metrics based on the attributes of these metrics or with combination of other metrics. Another great aspect of this visual is to analyze the selected variable with many metrics or attributes (dimensions) as the same time.

It’s not the type of visual I’d want to see on a dashboard, but I can see it as quite useful in exploratory data analysis.

Comments closed