The world is a fickle place.
Not quite a month ago I suggested that you don’t need to sweat the small stuff, and I meant it from the bottom of my not-shaped-like-a-pie-chart heart.
This month I’m going to suggest that you really do need to sweat the small stuff, and it comes from that very same chamber of sincerity.
I can get away with this of course by invoking clause 1 of the consultant handbook, paraphrased below:
…and if thine client pinneth thou down and demandeth an answer, thou shalt verily respond – It Depends.
Frustrating though it is to many people, there are very few answers that are right every time. It really does depend entirely on the question.
An example might help.
In my spare time I like to run. This keeps my oh-so-sincere heart in good shape, allows me to function better as a human being and, most importantly, gives me plenty of data to play around with.
One of my regular training sessions incorporates a tempo run. A typical tempo run would entail 20 minutes at an easy pace to warm up, 15 minutes at a fast pace and then 20 minutes to warm down. Here’s one I did recently, averaged up into those three distinct segments.
It looks like I nailed the brief – three distinct sections, warm up and warm down very similar and a quicker pace for the tempo portion.
But when you look at the detail, you get a totally different picture.
The data is still broken up into three phases for illustrative purposes but if view one gave us the duck floating on the water, this one takes a much better look at his legs.
The warm up and warm down phases are quite erratic, suggesting (quite correctly) that I had a lot of trouble locking into a consistent, slower pace than my normal run.
The tempo portion begins with the excited puppy phase at point 1, quickly followed by several moments of what are you doing…slow down culminating in the trough of disillusionment marked on point 2. An internal pep-talk leads to some classic peaks of overcompensation around point 3 and finally point 4 reveals the moment at which I started counting the seconds down until the torture could end.
Paying attention to the shape of your data often leads to the It Depends conundrum being resolved when deciding on the best way to represent it. With many types of data, revealing the shape can be critical to the analysis.
Here’s another running example taken straight from Garmin.
The numbers behind the data don’t really matter for this example – the key is the shape. Each of the eight peaks represents a hill sprint. The perfect data shape should be:
- Eight peaks of similar height (showing a consistent pace across all sprints)
- The distance between each peak is similar (showing a consistent recovery period)
- Each individual peak should gradually taper upwards (showing consistent, slightly increasing effort during the sprint)
The first three peaks in the example show a similar pattern to the excited puppy / disillusionment trough we looked at earlier. The remaining five show a much better overall pattern and are closer to the gradual upward taper we are looking for. The final peak clearly depicts the WIT (Woohoo It’s Over) effect as beer o’clock beckons (it was an afternoon workout, not a morning one honest…).
So, to summarise:
- Pay close attention to the shape of your data
- Think carefully about the question you are trying to answer
- …and always remember, it depends.
Oh, and if you still need some sound data advice then you could do worse than to chat to those nice folks at Synergy…
Connect with us!
Post written by Nik Eveleigh, Principal BI Consultant, Synergy.
Other articles in this series:
Making Healthy Data Choices – published September 2018
A helping hand(le) for your data – published October 2018
All roads (eventually) lead to data – published November 2018
The 12 Vizzies of Christmas – published December 2018
Don’t sweat the small stuff – published January 2019
Subscribe to our Monthly Blog
Image courtesy of Pixabay