Experimenting with Tableau: 13 Images which document my first time with my new favourite data visualisation software

I took an old Excel sheet that I worked on with the BBC last year and used it to experiment with Tableau. Here’s what happened…

Today I reintroduced myself to Tableau. In 2015, when I had just started my journey as a data journalist I was introduced to the program known as Tableau and was told of the wonders that it holds for a data journalist.

Back then, the sheer power it held scared me away from it. I was contempt with learning the basics first and using simple programs for my needs. After all, they worked fine and the saying, don’t fix what is not broken came to mind.

It’s also key to know that back then I was using very small data sets. Ones that could be researched, created and published in under 20 minutes. It was not until mid-2016 that I started to dabble in larger datasets and this is where I wish I had known Tableau sooner.

After spending the best part of my Sunday afternoon experimenting with the program, adding in my revenge porn data (that is a large and complex dataset) that I had worked on with the BBC and clicking random buttons until stuff worked, I developed a love for the program.

Throughout the experimentation, I took several screenshots and today I want to explore them with you.

Tableau is anything but useless and is something I regret waiting this long to fully get to know.

1: It’s very simple to get to know and uploading data is both quick and easy (Plus the formats they accept are great):

Uploading data was very simple (as you would expect) and took no time at all, even with a 15Mb file. Whilst I was not experimenting with Json (today at least) it offered the ability to upload in that file format as well as incorporate 3rd party API’s into the data offering an unlimited amount of possibility.

The screen is what you would expect (very Excel like) and to me, this was reassuring as it offered a ground that I was familiar with.

2: The magic happens in a window which is broken down really cleanly, making large data sets a breeze to handle:

Another thing I noticed was how easy it was to take the data from data/text form into a visual element. Having given up at this stage before in 2015, taking the time to mess around with it now allowed me to see just how powerful the program is.

With the clear naming of all my columns, it was easy to see what I had to do. The fact that all I had to do was drag and drop the data category into either the column or row data encouraged me that this would be much easier than I thought it would be.

3: Even if you don’t want to make charts, it’s great for analysing large data sets quickly:

One thing that I did come across in my experimentation with Tableau was that even if your goal is not to make a chart, the program is great at allowing you to analyse the data. When calculating large sums on excel it can often take time, but with Tableau, it was quick and efficient.

In my few hours with the program, I was able to try it out in a number of examples, but the best example was the one you see below. With just two datasets dragged from the left column, I was able to see that most police records did not record the age of victims of revenge porn. At the same time, I was also able to see 25 was the most common age, 11 was the youngest and 65 was the oldest.

It was experimenting with this that showcased to me the power that Tableau has, even when you’re not fully utilising the power of the software.

4: It’s easy to customise the chart to show the data in the way that you want to see it:

An additional point to the one above, it was very simple to change the layout of charts to be more suitable to what you want. You will notice that in this example the X axis has now changed from Index to Number of Records, allowing me to fully see the number of cases for each age group. A simple and easy feature was easy to change to suit your needs.

Aside from that, you will notice in the right column several highlighted and greyed boxes. This is where Tableau is really great. Greyed out ones won’t work (because the data is not compatible) whilst the highlighted ones will, when selected, change the data to show the visual aid allowing you to quickly mess around with the data to find the best visual for your data.

5: It did have some weird moments where the graphs went weird for no apparent reason:

There were some times when the experimentation went all a bit wrong, see the example below which was meant to be a pie chart, but for reasons unknown, resulted in looking anything like one.

What went wrong? Well, it turns out I had changed the size of the pie-chart and accidentally made it a square instead of allowing a responsive layout change when I made it bigger. I only learnt this because of the great inbuilt support that Tableau has (videos, chat, forums) but it was an interesting lesson to learn.

6: Even by putting random data categories in the columns and rows, I found useful information:

One graph mode was called automatic and whilst it was not the best visual tool, it did really help present the data.

In this case, I was looking at the age and comparing the police forces, which even in its table form allows me to quickly glance at the data. Not a key point, but one that my experimentation showed-up (by mistake, I was just pressing buttons) and lucky, one that I’m glad I found.

7: The one thing I did learn for my experimentation, is that it’s really just Microsoft Excel (except much better):

Again, not a major point, but I found that Tableau was able to present the data really quickly. Compared to running it on Excel (Which with this much data took a while on my Surface) the results were given to me in next to no time.

One of the biggest things I can take away from experimentation is just how stupid I was taking so long to fully get into this program. I always saw it as just a data visual tool, but actually, it’s just as good (If not better) to analysis data. One particular element I really see working is using the program to search a large database to gather, quickly the key points in any data set without having to remember a million and one different =sum formulas.

8: I learnt so much more (about revenge porn) in the last 5 hours then I did when writing and researching the story:

This story was a big break for me, the first one that the BBC (with the help of Paul Bradshaw) used and showcased to me just how much fun and rewarding data journalism can be.

However, despite searching the Excel sheet for countless hours in 2015, two years later running the same sheet through Tableau is highlighting new story elements that I did not know before and almost wish I had known.

9: Oh and I did learn (after trying several times) how to make Pie charts, not that it still did not have issues:

After watching a YouTube video (there are several dozen for Tableau, each is great and teach you everything that you would want and/or need to know) I was able to fix the pie-chart problem earlier. The issue turned out to be that I had selected the wrong view (see tab in the second bar) and needed to select entire view to fully see it.

The great thing about experimentation is that you can make a mistake and later learn how to fix it. This was one of those cases.

One thing that I did notice was that for all the good Tableau has, it can’t avoid ground-based problems that every data visualisation has, in this case too much data. As you can see the problem with the chart (even before adding tags e.t.c) is that there is a lot of data which would make it hard for any viewer to read.

10: Some graphs, whilst fascinating to look at from an analysis perspective are completely useless to the reader:

I found this one truly by mistake and whilst it did not have any practical applications due to the size issue, it did showcase a new presentation method that has pipped my interest and will result in my visiting this method again in the future.

It was interesting to break down such a large amount of data into so many pie-charts and seeing in this case, if the victim's abuser was an ex-partner what age was this most common (in this case it was 24–34) — It’s an interesting element that adds to the story that the numbers tell and something that I missed in my original search of the data. Once again, showing how easy it is to miss something when searching such a large database.

11: If you want to know it, the changes are Tableau will allow you to visualise it, even if it does not look good:

I pressed a button by mistake and ended up with this beauty, which whilst very long, is now my favourite graph. Showing that for those published FOI’s which show if the abuser was the victim's ex-partner that 19 was the most common age. It was an interesting mistake which helped me discover another interesting way you can combine to data ‘dimensions’ to produce interesting, revealing and useful data visuals.

12: You can change the visuals really quickly. Allowing you to do rankings at a fingers click:

My favourite trick is that by clicking the box that appears in the top left corner (blue, highlighted) you can select a button which will change the data set to appear as you see from largest value to smallest (or vice-versa if you wish).

I love this feature, just because it means there is less work for you to do. I’ve spent countless hours on Infogr.am and Excel having to reorder data e.t.c so that when I upload it, the data will appear in order.

13: New graphs are fun to explore and showcase some new tools for some old methods:

However, despite everything that I mentioned in the previous 12 points, by far my favourite feature is that Tableau introduced me to some great new graph types that I have never seen before.

The bubble chart below is just one example. Now while it shows what you would expect (as those forces mentioned are the biggest in the country) it provides a new an interesting way for users to see that data instead of just a normal bar/line or pie-chart.