#FastFriday - Number 1

Welcome to the first in the #FastFriday series.

For context, this is a weekly challenge that Ben Moss, Simona Loffredo and myself will be participating in. The concept is as follows - We have to create a visualisation created from a dataset which we see for the first time  when we connect to Tableau. We're starting with a time limit of SIX (6) minutes to create a visualisation with meaning.

The objective? To learn about interpreting data better. To understand each other's approach when faced with a new data set.

And how well we viz under pressure.

We will screen record and then blog about our experiences.

Make sense? Check out my video below (complete with my mutterings while I build)

TIME: 6m 21s

The beginning

First of all, I scrolled across the data in the data source window - to get a feel of the type of data, and the width (number of columns). Following that, I went to the sheet and immediately started looking at the measures and dimensions, using Number of records to better understand the dataset.

I think this worked a bit as I understood that the row level was at 'Data Zone' level, and then I saw the hierarchy. When working with data longer term, I think this sort of exploratory analysis is really important - And further, that knowing the relationships within the data help you to understand if there's any more data prep/cleaning/ETL (extract, transform, load) to be done.

You can see that I couldn't see the hierarchy at all with number of records so dragged it off and went to a more tabular view by pivoting.

The Middle

Wow. So many measures! With wide field names, and given my jittery state, it took me a while to comprehend what measures were there, and what I could show with this data (as you can hear) - I settled on population, created a bar chart with the label inside the bar (which admittedly, ate a bit of time) and then left it. This entire process, took about half the time. So a bit of panic began to set in...

Processing the field names was interesting, because in the time-pressured state, you have to really be careful what you're building, and how the story will layout. The scattergun approach to data-viz where you conceptulise then create all gets jumbled into one slick process. I say slick. I mean more like Kermit the Frog running around on stage with his arms in the air.

The Dilemma

The scatter plot was my plan when I decided to use the bar as a filter - this would mean that it would be exploratory but with some takeaways with the original view (I hoped!)

I started messing around with some measures, eventually deciding i should actually read the measure names rather than mindlessly dragging and dropping.

I'm not sure exactly what I was trying to show with the scatter plot, with income deprived folk on the x-axis and education level on the Y, but there was a trend and i thought it would be interesting to see clusters by the local authority (or something to that effect)

Messing around with colours and trend lines ate a bit of time, but i thought I'd at at least a bit of colour here..

The biggest mistake? Thinking I had enough time to add summary cards. If I had a clear idea of how this would play out and how I'd do this (order of dragging and speedy formatting) I would have probably done it for sure. But in this case, I didn't - so it just ate into time I could have spent on the dashboard, or for tweaking the other visualisations. Summary cards look cool, but with the time constraints it was perhaps not the best use of my time.

Always learning...

The Grand Finale

I love the speed that you can do things in Tableau - no less that I can create my dashboard in two clicks... Four if you include the fact I brought in my dud sheet! But with a few clicks and a title, I had a semi interesting dashboard (with a bit of a story? Kinda?)

This was intense man. Like, I didn't expect to spend this long working through the dashboard and developing something on the fly. Watching back the part where I fumble around the cards, is interesting as there's a point if I was watching where I think if I was watching someone, I'd feel they wasted time there. So it's a weird feeling, but one which I thoroughly enjoyed. 

Then - the most rewarding part? Watching Simona and Ben's entries. All three of us approach data in such different ways, it's fascinating! The fact that Ben immediately planned his dashboard, and that Simona took the gamble of a) using a calculated field and b) ACTUALLY NAMING IT was so rogue. Interestingly, they both used measure names/values - which was an 'ahhhhhh, I totally should have done that' moment for me.

Check out their efforts here:

Ben Moss: http://benjnmoss.wordpress.com/2016/05/27/fastfriday-6-minute-vizzes-1/

Simona Loffredo: http://thevizconnoisseur.wordpress.com/2016/05/27/fastfriday-series-first-episode/

Want to get involved? Here's a fresh dataset for you to try!

Remember - Time yourself for 6 minutes AND no peeking beforehand! (click here)

Read More

Google Data Studio - First thoughts

As you all are aware, in the last year I've spent almost every single day using Tableau. However, I've done my best to try out other data visualisation tools. In the past, I've reviewed QlikSense and Power BI - So of course, when I found out that Google had released their own data visualisation tool, I had to check it out.

The Lowdown

The software is called 'Google Data Studio' (https://datastudio.google.com/#/navigation/reporting) and it is currently still in beta.

The paid version (Data Studio 360) is/will be a part of Google Analytics 360 Suite - Which also includes

• Analytics 360
• Tag Manager 360
• Optimize 360 (also in beta)
• Attribution 360
• Audience Centre (also in beta)

Check out the full range here (https://www.google.co.uk/analytics/360-suite/#?modal_active=none) - Of course, being Google, this lends itself to the 'wider' Google stack; following the 'one login, full integration' sort of model. In order to purchase the Analytics 360 Suite (or, I'd imagine you have to buy the whole package) you have to either contact Google, or one of Google's partner resellers.

BUT! A couple of days ago, Google announced a free version of Data Studio, aimed at individuals and smaller teams. The limitations? You can only create 5 reports (360 allows unlimited) & you can primarily connect to Google Sheets, BigQuery, AdWords, Google Analytics, YouTube and Attribution 360. Though it's currently focussed on the Google 'range', there is talk that SQL database connectors will be rolled out.

Data Studio's goal is to 'help you accomplish the three primary tasks you need to create beautiful reports that tell beautiful stories with your data'

Connect. Visualise. Share.

So very Google!

So, one of the immediate main drawbacks for me (as I write this on the train) is that the product does not work offline - Which makes sense to an extent; Google Docs etc are all browser based, as is the Chromebook. So it's not an uncommon thing to see. I'd imagine that the paid version *does* have an offline mode, but of course I hope so anyway (I don't know this as a fact)

Pre-demo thoughts

Instead of testing the tool (which I will do later), let me talk about first thoughts in this post. Watching the 'demo' video was slick. As it is Google, there's a well narrated video with beautiful visuals and a nice UX... Which made me feel a bit concerned as a huge Tableau practitioner!

Connecting to data looks simple enough, but as I mentioned before, I seems to be very Google specific (for now) - Which isn't an issue really, as most data can be in Excel formats etc.

The interface itself is very similar to what people who have extensively used Google Products online, the sort've format of Google Docs/Slides/Sheets etc. And of course, Google champion themselves for the simultaneous collaboration - which is possible within this tool (though you're able to set permissions) - Check out the image (taken from the Google Data Studio website, but annotated by me)

This is the sort've 'opening view' of Data Studio in such a Google way and then once you'd opened the datasource, the building window looks a bit like this

When I saw this, I thought two things; one, that the interface is very similar to QlikSense (config pane on the right, chart controls on the top) but second; green things, blue things. This refers back to the post written by Tom Brown years ago which I always recommend to people starting with Tableau to understand continuous vs discrete - and Google have done something very similar with the 'Pills' that the user can bring in.

So this is a really basic overview and first thoughts, from reading documentation, watching the video and seeing how the interface looks without actually playing with it.

Conclusions - For now

Moving forward, I'm going to try and have a proper play than this very basic overview - but my initial thoughts are fairly positive. It may not have the deeper analysis tools which Tableau has, including the level of customisability - but for firms/individuals who are doing very basic reporting, this doesn't seem a bad place to start.

To find out more, check out https://www.google.co.uk/analytics/360-suite/data-studio/ for more on this.

To try it out (and certainly where I'm going to start) check out - https://datastudio.google.com/#/reporting/0B_U5RNpwhcE6R3d5Nk5zOE5ZSGM

Unfortunately, at the time of writing Google Data Studio isn't available in the UK. I'll dig a bit deeper into the product once it is available here.

Read More

Creating a Soundwave Chart in Tableau

EDIT: This post was previously named "How to make the histogram which looks like a audio wave thing (in Tableau!)" - But that seemed a bit naff. So I updated the title.
And added a viz at the bottom!

I use sports data a lot - I understand it intuitively and always have more questions I want to ask of it and find out more. So you could say I think about it fairly frequently, even if it's a subconscious thing.

Such an event occured this morning as I was walking to work from Liverpool Street station. One of the visualisations I'd created yesterday took two different line charts and filled the space in between. It was a crude solution, but it worked and it got me thinking... Remember the shot worm (see below)? Could I recreate something like that in Tableau?

This is from Peter Beshai's awesome interactive dashboard 'Buckets'

Luckily, I had a basketball dataset to hand, as one of the Makeover Monday (run by Andy Kriebel and Andy Cotgreave) challenges were on Steph Curry and his ridiculous streak.

Long story short, I did a mini-makeover of my own effort, and ended up with the below.

NB; it's very crude, I didn't update the headers or annotations or anything.. I even moved Steph across in Paint.


How was this done? Check out the short video below.

UPDATE (31/05): I actually created a visualisation which uses this technique to show how the chart could be used with NBA Statistics. Check it out below.

Read More

13th of May '15 | One of the best days of my life

Some things never change

Let's wind the clock back a year.

I am on a train back from London (of course), travelling through the beautiful country side, hurtling back to the South-West, to Plymouth where I studied my undergraduate degree. There's a copy of "Environmental and Resource Economics" by Tom Tietenberg and Lynne Lewis open in front of me, but I'm not paying it much attention.

The 13th of May 2015 was perhaps one of the best days of my life so far. And a lot of it I think has to do with luck and opportunity.

Setting the scene

I studied Economics for four years and took a sandwich year in between where I worked between my second and third year - this is when I fell in love with the idea of working with data and presenting it back with added analysis. 

I decided I wanted to work in data visualisation toward the end of 2014, and when I read Neil Charles' post on 'How to do football analysis in Tableau' in Febuary 2015, I think that changed everything.

It was through NeilCharles and Steve Fenn I became interested in Tableau, but one opportune tweetwhich I read on the way home by Chris Love (who I followed after seeing his amazing Iron Viz entry on the History of the Football League) which was the largest of what I call the bizarre series of events which has lead me to end up where I am.

Yes, a year ago today, I interviewed to work at the Information Lab and be part of the inaugural class of the Data School.

8 hours for 30 minutes - Worth it?

I'm sure you've heard of the process, sending a Tableau Public viz, having a chat with Andy Kriebel for 30 minutes about it and about myself - God, looking back at that visualisation, I can pick SO many holes in it.

Click!

Look at it. An OK effort for the first time I was using Tableau for something other than football data, but so many issues.

• The years are along the y-axis (I honestly have no idea why I did this)
• There's clear red-green colourblindness issues
• The stacked bar is kinda hideous

Stepping back even further, here's the first visualisation I posted on Tableau Public.

Sized automatic. Huge right hand panel. Cool trend line. (Click!)

It's using TransferMarkt data which I scraped using Import.io, but he's a scatter plot - I distinctly remember being impressed that I could actually add annotations and comment boxes. Bless.

The Dilemma

The rub for me was that, after passing the phone interview, the date given to us by Andy was slap bang in the middle of my exams. These were my finals, the four exams I needed to pass to walk out with a credible degree. Tuesday 12th, I had a Game Theory exam. Thursday 14th I had an Environmental Economics exam. And here was my prospective employer, a Zen Master no less, asking me if I can make the Wednesday work (I did ask to have it rescheduled - the answer was "are you sure you can't make it work?")

I gambled. The 4 hours there and 4 hours back was enough time to revise on the train, I could do some reading, check my notes. And hey, if I'm not prepared now, there's no hope right?

The interview process is akin to speed dating - 15 minutes in a competency based interview, 15 minutes presenting my work. Then told "we'll let you know" and poof. I'm wandering down Tottenham Court Road with no idea how the interview had gone.

Meanwhile, back on the train..

And there I was - hurtling through the South-West, mind wandering away from willingness to pay models and carbon tariffs, I receive 3 emails.

The first from my lecturer and personal tutor, who asks how the interview went and informs me of my final coursework grade - 89%! I was dumbfounded - I didn't expect that! 

The second, an automated one - telling me my dissertation is ready for collection - and I was overjoyed with my result of 69.1%

And finally, the pièce de résistance -  A call from Andy offering my a job.
Incredible scenes. My feelings are fairly accurately depicted in the above .gif

Reflections

I cannot imagine a better job right now - I enjoy what I do, I'm driven to continue learning... There was a great moment in our third or fourth week where I stopped, and realised that I was now being paid to do something I was doing in my spare time. Simply amazing.

And the best part?

It is so rewarding to be able to see the progress.

A great example is the visualisation below - I think it's among my favourites I've created (along with a couple of the client projects I've done & the tiled map I created in Tableau 10 last week)

The best bit? The data used here is almost identical to that first scatter plot. I cannot wait to see what else I can learn in another 12 months.

Click!

So, I think the only thing to do is this - Thank you so much to the Information Lab, to Tom, to Andy and to all of my fellow Data Schoolers for being simply fantastic. And last, but never ever least - The Tableau Community. It never ceases to amaze me how supportive, constructive and collaborative this group of software lovers are. It's simply amazing, inspiring & I thank you all.

Let's see what happens between today and the 13th of May 2017!

Watch. This. Space.

Read More

Regarding 'Chart Junk'

Graphical elegance is often found in simplicity of design and complexity of data. 

The visual/functional spectrum is something that's grown and grown as an interest for me, particularly after trawling through the blogs and comments on Stephen Few's blog (the ones on McCandless are incredibly interesting, I found this recent one extremely insightful) that there was a clear divide among people who 'practice' data visualisation (let's not even get into the 'is data viz a science or an art' debate) - Should your chart/visualisation serve the purpose to inform, or merely attract interest? Infographics and KPI Dashboards are two examples at either end of this spectrum.

It's very easy to take the numerous amount of literature, blog posts and advice from experts as gospel - you can enforce a bar chart on most data types, and everyone seems to swear that pie charts are evil (well, almost everyone - Matt Francis has kicked off a website to debunk data viz myths) - but falling down deeper and deeper into this rabbit hole has lead me to question things a lot more. Why are thing referred to as chart junk? And most importantly, what IS chart junk?

Chartjunk | Refers to all visual elements in charts and graphs that are not necessary to comprehend the information represented on the graph, or that distract the viewer from this information (Wikipedia, inspired heavily by Tufte)

I feel like there's a bit more to the term chartjunk than this. Tufte mainly talks about the visual elements - which refers to 'bad' use of colour, shading, borders etc. The 'unnecessary use of ink' is seen to be distracting, and adds nothing - how much 'data ink' can you delete before the visualization stops making sense?

The above chart was found here on Bloomberg Businessweek, where they explored some Tufte-isms - Including needless colour coding, needless visual elements such as gridlines, 3D effects and labeling, and simply anything distracting from the visualization as a whole.

Essentially, anything non-informative or elements of the chart which aren't adding anything to the visualization qualifies - added dimensions, questionable use of colour, visual format elements like grid-lines not adding any more information for the end user. While reading up for this post, it was fascinating to read Tufte on this subject - the man is *so* passionate! Take these choice parts (from here);

"The interior decoration of graphics generates a lot of ink that does not tell the viewer anything new... Like weeds, many varieties of chartjunk flourish"

But for me, chartjunk can also refer to bad chart decisions - and this is actually something I've been guilty of during 'trials' of charts and dashboards.

Let's start with the dashboard; with 'effective dashboard design', depending on the message that you're trying to convey using the visualization, each part of the dashboard should contribute something - otherwise, it can count as 'chartjunk'.

- How much value is that bar chart adding?

- Is that trend line showing the same data differently?

- Why are there packed bubbles used here?

The principle, however, remains the same - How much can you erase/delete before the objective or purpose of the dashboard (or visualisation) becomes redundant? Robert Kosara has a really interesting blog post on chartjunk (here) which links to Stephen Few (here) - The crux being that visual elements and use of ink aren't bad 'period'.. But more that they should be used with care, and the end user in mind.

For the way that 3D charts/shapes, shading etc work - fine. But for the use of colour, there's a whole different spectrum to consider; but with relation to chartjunk, and taking it at the face value of 'usage of data ink' - I tend to agree with Few and Kosara... It depends, therefore 'relative' chartjunk has it's place.

The key part of gridlines and strong colours is the impact; by this, it refers to weight of a line chart, or the gradient used for a bar chart or the strength of the colour of the gridlines. Depending on the end user, gridlines are  useful for allowing the eye to follow, and shading highlights certain areas - and extreme shading (also known as highlighting..) brings out the strongest elements of a visualization, the parts which carry the message or that the end user should be drawn to.

Visual elements can help guide the user, and ensuring that there's enough guidance, given that context in visualization is key - a good question to ask yourself is, where does this viz sit in the wider scheme of things? How does it support or add to the analysis/question posed?

I think the term 'chartjunk' has evolved and developed, but the ever poignant statement of 'it depends...' remains over it's head. As always, I would love to hear your thoughts on this topic!

Read More

Defining the #TableauShuffle

I've now used Tableau almost every day since June, and even now I'm impressed at how much thinking it does for me - And in particular, I'm talking about the speed at which queries are run, and how this allows the user to iterate quickly* (*quickly here coming with a caveat of the structure, size and nature of the data you are connecting to Tableau..) "at the speed of thought" - Meaning you're able to answer questions quickly, easily and actually, pretty effectively.

How good are you at Powerpoint? 
I excel. 

Was that a Microsoft Office joke?
Word.

When asked where Tableau falls in the Powerpoint/Excel spectrum (a remarkably common question), I launch into a anecdote (obviously) of when I was intern during my year in industry working in business development. I'd work on a 2-3 month project, collating qualitative and quantitative data before presenting it in a 3-4 hour meeting to the project leaders/clients. My worst nightmare? Being stopped on slide 36 of 60 and being asked to change the colour of a chart, or to pivot it, or to add another chart underneath it to show something in a different way. Argh! Such a small pain, but meaning another month attached before it would disappear off my plate.

This story has seemed to resonate with quite a few analysts I've met since I started with the Information Lab - The iteration process, people not trusting the data they're seeing & the painful cost of time. Wouldn't it be nice if there was a simpler way?

Enter Tableau - And more importantly, the #TableauShuffle.

I have to credit Craig Bloodworth with the coinage of this term which I have used endlessly since he challenge us (#DS1) to think about what's going to happen to the visualisation before we drag and drop a field onto columns or rows. This is where I began to understand the value of actually understanding the tool.

So what is the Tableau Shuffle?

(apart from a super cool move which should be added to the range of dance moves that Jewel Loree came up with for #TC15) (Jewel, if you're reading this - if this ever gets repeated, it would make my life if the Tableau Shuffle was added to #WatchMeViz!)

Though, perhaps thie drag and drop move works quite well...

The Tableau Shuffle is the iterative process of dragging and dropping dimensions and measures to see what happens - It's one of the best parts of Tableau. Indeed, I didn't know that double-clicking a field would drop it on rows and columns until I started at the Data School. And it was a few weeks before I discovered right click and drag - But I digress...

For example...
I used the TransferMarkt dataset I uploaded a few weeks ago (the first viz I made for this is here) to do this (click the image to bigger it);

Here, I'm looking at the age curve of players who start a match across the entire dataset - As we can see, there's a clear bell curve. Now I see this, I like it, I understand it - But I want to know how does this compare to the number of sub appearances?

Nice. But what about by league?

Yup. That's it.

This is what I love about the Tableau Shuffle. You can drag something on, dislike it, take it off - Drag something else on, change the chart type.

*Everyday I'm shuf-shufflin*

And it just works.

A step further to this was the last challenge I set for the 2nd Cohort of the Data School (#DS2) in the Data School Gym (where we flex our viz muscles of course) - The premise was using this button 

which turns off automatic updates... And then building a dashboard! It's harder than you think, and makes you think about the composition of a view while you're building it.

I'd love to know you guys' thoughts on the Tableau Shuffle - And any experiences you've had where you would have loved to use such a technique.

Read More

About Mick, Marcus & some things on analytics

This is a repost of a long post I wrote on the Ipswich fanzine 'TWTD' in the forum

A caveat before I begin; been ages since I posted on Twitter, and am far more active on Twitter - but a conversation or two over the last few weeks (and ITFC Twitter being overrun by crappy generic 'fan' accounts), I'm back here to get some long-read views out & maybe start a discussion which invariably will go off on a tangent.

First; Mick McCarthy - This season is the plateau where many are seeing how so much came good at once last season (albeit less so in the second half). I appreciate the functional style of play, and note that Leicester do it similarly (albeit, so much better) but this season has been a bit meh - so much, that I've not blogged about their #numbers all year - and probably won't until the end of the season now.

What to do if 'MM leaves for Villa'? Let him - He's a good bloke, knows his skills and limitations, and for me the last few games have echoed his time with Wolves in the Championship. The draw specialists, who circled around mid-table for a while before hitting it right. If he doesn't leave, in my eyes he has two transfer windows before parting ways at the end of next season - Barring any categorical disaster or success.

The main issue for me is this; The state of a) the squad and b) the 'philosophy' that is employed by Ipswich Town FC is tailored to these players by emphasising hard work, specific shooters and a firm back line. This year, Town have had issues with injury, but there hasn't been a single XI who have backed themselves as THE best 11. And this is a shame really.

So, MM leaves and Manager X comes in - The state of this squad means ME should either hire someone similar to Mick, or slightly more modern than him in terms of style of play and type of player. This takes us to the train of thought of a 'transition' manager before settling on a new model of success after the transition has been completed.

And all of this takes the biggest commodity in football - Time. The disparity between the PL and Championship is growing year by year, and it's becoming tougher and tougher to compete. But, shy of taking a huge gamble, risking relegation & pressing the reset button like our East Anglian neighbours, I don't see any solution other than the two options; a go between, or a gamble.

I touted Nigel Pearson last night, not the most fashionable choice, but I was lucky enough to meet a couple of the Leicester analysts at the Opta forum, and the latter half of last season was Pearson drilling the team to work hard, power through and ensure survival - And it struck me to be akin to the MM model. Again, the caveat remains it would be a 2 year deal where he would move the club toward a more affable style of play, but this period for me would be key. First, to introduce and develop academy prospects and second, for the internal structure of the club.

I think Ipswich should move away from the white, British, between 45 and 55 years old with some sort of historic success in the Championship approach, and implement a proper strategy and a way of thinking to help advance the club. As some of you may be aware, I'm a fan of the anaytical approach but one which bases itself on championing coach knowledge to be combined with additional insights that data can provide. Average player positioning, on-the-ball actions & development of tactics, styles and such are all possible in conjunction with the right amount of data. And further, having the option of academy data really can help track progress of a player & further understanding. In my eyes, there's so much that Ipswich could do better here - starting with *actually* having a presence in this at the Performance Analyst level, beyond simple video analysis & scoreboard stats. Norwich, believe it or not, have a remarkably structured platform for this, and are doing some annoyingly smart things with both data & their academy development. (And I wrote to Ipswich three summers in a row, asking to propose something/help in this instance with no reply - probably asking the wrong people)

So; my three take aways;

1 - If MM leaves, Ipswich have options but it would leave us in a lurch as the squad is neither here nor there at this point

2 - When the next manager comes in, the choice for ME is to either be radical with a new coach, a new coach + DoF, or a transitional coach while developing a new structure

3 - Modernise backroom/club activity to keep up with the modern age

Thanks for reading! Thoughts welcome.

Read More

#TableauKISS - Lessons from i, Robot

I loved studying Economics; it's a social science, where it studies the behaviour of society, individuals and 'economies' as a whole, but then it also attempts to model, quantify and segment the vast amount of variables and factors which are always in flux. This allowed us as students to choose which we preferred, and which school of thought we wanted to subscribe to.

In some ways, I hated this about economics - it could be explained in such layman terms, and concepts could be discussed with great vigour. But then someone could equally refer to a mathematical model, or a quantification of impact, or use Shapely values to place worth on something to take the discussion in a different, arguably more scientific direction. I liked my economics simple, understandable and as a trigger for thoughts, debate and discussion - in a way which could engage as many people as possible.

Tonight, Chris Love launched his #TableauKISS movement - taking Tableau back to it's basics to highlight what can be achieved without edge cases, hacks or advanced charts, or charts which aren't native to Tableau.

On Friday, I was able to demonstrate Tableau to a room full of Excel and Powerpoint junkies - I was promised 30 minutes to talk through it but ended up being given 7 - and in that 7 minutes, the energy and excitement that I was able to generate was something which I didn't expect.

Did I create a Marimekko chart? A Sankey chart? A chord diagram? No - four charts; a bar chart, a line chart, a scatter plot and a map. Put them all on a dashboard, add some filters and suddenly I'd created something that the room had never thought would be possible.

Trying to display everything I knew in that demo, or attempting to replicate a chord diagram, a Sankey chart or any other incredibly amazing chart type that those who have mastered Tableau and *just get it* do - It's cool, yes - but more often than not, the message can be clearer and the information can be portrayed in a simpler way than getting lost in calculations and hacks.

The beauty of Tableau is that it allows anyone, everyone to pick up some data and begin creating something amazing in a quick, informative and beautiful way - that's it. And pushing Tableau to it's limit is simply awe-inspiring - see Josh Milligan's games, Noah Salvaterra's 3D Donut chart (I love this one) and the recent gorgeous Beatles viz by Adam McCann to just name 3 incredible visualisations. But with no coding knowledge, virtually no software or database experience and a passion for numbers in sport - I'd never have been able to do this.

This notion has struck me often over the past week, especially following Chris' post about his love/hate relationship with Tableau edge-cases - And it took me back to my four months as a Data School padawan.

If I haven't mentioned it enough, the Data School was an incredible experience where we spend virtually every day from June until November using Tableau and Alteryx, learning more and honing our data visualisation skills. 

But I remember a few weeks after we learnt *that* much Tableau in *that* short a space of time, that I found myself trying to apply everything I'd ever learnt in Tableau on every week-long project. And it always came back to the same thing - this works, but was it the right choice? What was good? What could have been better?

Some of the best Tableau visualisations are just absolutely simple, with a clear message - and when I was mulling over this idea, a particularly poignant quote from i, Robot came to me.

This was the one key thing for me, and perhaps a turning point - Asking the right questions. When you learn a whole load of tips, tricks and lessons in using Tableau, understanding data visualisation and telling stories through data... it's incredibly easy to try and apply all the tips and tricks at once to show that you *can* do it all. And I definitely have fallen foul of chart junk before being shaken to reality.

But as the weeks went by, I felt that I - and also the others around me at the Data School - found myself thinking two or three steps ahead. What is the key message I want this visualisation to portray? Should it be exploratory or explanatory? What will add to the visualisation, and what will take it away?

This preset of questions allowed me to have a repository of ideas to pull from - and the weirdest part? Most of them were remarkably simple chart types. Bars, lines, scatter plots, maps. Things that Tableau almost gifts to the users on a plate. I appreciated that learning Tableau every day meant that I had covered an unimaginable amount of content - but once I started to take my toolbox, take a step back and actually think, plan and start at the base of the problem I wanted to solve or question I wanted to answer.. The solution largely lay in a really simple idea.

I'm always learning - and I don't ever want to stop, but understanding that complex visualisations are just that, and appreciating the time that they take to create puts things into perspective. Personally, I'm a big fan of the #TableauKISS movement, and I think it's a great idea that I'm hoping to contribute to as often as possible in the next few weeks. 

Read More

Makeover Monday - FiveThirtyEight

I've agreed to partake in the Makeover Monday series which Andy Kriebel and Cotgreave will be doing, which hopefully will encourage me to work with different datasets as well as hone the way I think about data visualisation as a whole. 

If you're also interested in the project, check out Andy's blog post about it here

Accepting this challenge, I took it one step further and also attempted to 'blog' my approach to this week's makeover using story points - I've not used story points at all really, and recently was tweeting wondering what ideas people had for good uses.

I've come to explain the pro's of using a Story Points system over standalone dashboards and tabbing across as Story Points allowing the user to choose one dashboard and cycle through it, highlighting different aspects as the presenter, well, tells a story.

I think this is where Story Points can be powerful - However, I think there's definitely aspects of it which can be honed and amended to ensure that people;

a) use it properly and
b) are able to harness the full power of Story Points.

Andy's post challenges everyone to see what they can do in just an hour; I'll admit I spent just over 2 on this - with about 20-30 more minutes tweaking and then setting up the story points. The formatting itself was the part which took the longest - tidying up tooltips, making sure none of the visuals were overlapping or intruding upon one and other.

I'd love to hear and get some feedback on this makeover - Hopefully the first of many to come!
Oh, a sidenote - I know jack all about baseball.

Read More

Different types of Attacking-Midfielders; a thought starter

Disclaimer; this post isn't about Ozil. Much.

Mesut Ozil is killing it this season.

With over 15 assists already, the German international is on course to hit and surpass Thierry Henry's record of 20 assists in a Premier League season (anyone else unsurprised that this is another Arsenal player under Wenger who holds this accolade?) - But what shouldn't be forgotten is the stick that Arsenal's record signing got in the first season and a half he spent in North London. As the narrative goes, Mesut was not a big game player nor was he the type to be prolific in front of goal.


This tweet by Simon Brundish got me thinking; how long should different types of players be given to make an 'impact'?

Let's examine another player alongside Ozil; Alexis Sanchez. Arguably, he also counts as an attacking-midfielder and I'm sure it cannot be argued that his debut Premier League season was better than Ozil's.

Where Ozil played 2150 minutes in the Premier League, scored 5 goals and registered 9 assists (a scoring contribution of 0.586 per 90) in his first season, Sanchez played 2953 minutes, scoring 16 goals and 8 assists (a scoring contribution of 0.732 per 90)

So Alexis had a higher impact; but ultimately, they are two very different attacking midfielders. Where Ozil is a 'creator' and a orchastrator, Alexis is perhaps more direct. Let's see if this hypothesis holds.

Alexis took 3.5 shots per game, made 2.3 key passes and attempted 3.3 dribbles per game. Conversely, Mesut's first season yielded 1.2 shots per game, 2.9 key passes and 1.7 dribbles.

Already we can see Ozil's emphasis on creating chances than attacking a full back, and perhaps this is why he tended to play narrow for Arsenal even when placed on the wing.

But the secondary factor is this; where Alexis' game relies upon getting the ball, attacking individually, receiving and letting shots fly, Ozil needs to have the movement, space and time to pick out the passes he wants to make. And this would require the players around him to also understand his game.

The purpose of this post is Liverpool's #11, Roberto Firmino. As a rule of thumb (which stems from probably a comfort with multiples of 5), pundits refer to a great midfield player as someone who posts 10 goals and 10 assists a season.

In Ozil's three seasons with Real, he averaged about 6 goals and 15 assists per season.

Alexis at Barca in his three La Liga seasons averaged 13 goals and 8 assists per season.

So where does this leave Bobby Firmino? In his last 3 seasons at Hoffenheim, he averaged 9 goals and 8 assists a season - so perhaps a bit more balanced than the examples examined above. Last season, he took 2.9 shots per game, made 2.1 key passes and 4.2 dribbles - which actually points him to be closer to Sanchez than Ozil in this comparison - his scoring contribution was (in 2918 minutes played) 0.524, with 7 goals and 10 assists. These came in 33 starts, and was slightly down from his previous season. Another factor to his slow start at Anfield could be starting every game two seasons back to back, and representing Brazil at the Copa America this summer.But I am no sports scientist and cannot speculate on this in an informed manner.



 Using just one example of comparing Ozil and Alexis (given their 3 season stints in Spain prior to joining Arsenal), this could be a reason why Firmino is taking time to bed into Liverpool.

From my understanding, attacking midfielders can be 'a Sanchez'; a direct runner, who takes a lot of shots, is good with the ball at their feet. Or, they could be more of 'an Ozil'; less shots, more creativity, with an emphasis on assisting than scoring. The latter I feel requires a few things.

• Understanding the role; making sure that the role and all the tasks a player has to complete on the pitch are understood and being met - and that they are comfortable doing so
• Cohesion with teammates; for Ozil, players need to trust that if they make the run, Ozil can find them, and similarly that if they are to utilise his creativity, the man needs space. This positional understanding, trust & chemistry are important to ensure moves look as fluid as they can.
• Acclimatising mentally & physically; this includes on and off the pitch. On the pitch, the pace, space and movement need to be adapted for the environment an attacking midfielder finds themselves, and off the pitch, the club should ensure that the player is in the right state of mind to perform and 'hit the ground running'

In this vein, I feel that though Firmino may not have sparkled, his passing networks and interplay with his team-mates will increase similar to Ozil's has, and he possesses the ability to run at full-backs and contribute in a way that Sanchez can. It'll be interesting to see how Klopp utilises and changes the use of the Brazilian midfielder.

Read More