QS3 - Sleeping in December

For the month of December, I looked at my sleeping patterns. Since about November 2015, I've worn a Fitbit most days. I own a Fitbit HR, so my sleep is automatically logged. Now, one small thing that annoys me about Fitbit - actually, it's kinda big, but I hate the fact you can't easily access ALL of your data. Of course, this is probably possible through an IFTTT recipe, or accessing the API to get the data you need - but things like my heart rate, or my hour by hour steps... I can't easily download that! Like, what the hell Fitbit - This is my data!
</end rant>

So - I downloaded a months worth of sleeping data from Fitbit's website, and connected it to Tableau. I spent a good amount of time on the #TableauShuffle stage with this month's data. I couldn't decide whether I wanted a connected scatter plot, a Gantt chart.. In fact, let me show you what the connected scatter looked like..

It's a bit of a mess, with no real trend! I guess I didn't choose the moniker of "scribblr" for nothing...

Anyway, as a side note, sleep efficiency is a calculation which Fitbit define as..

"We use the following equation to calculate your sleep efficiency:
100 * time asleep 
/
(time asleep + time restless + time awoken during sleep)"

It was an additional metric I added as it's on my Fitbit dashboard.

This project almost got to a stage where I was just going to publish the classic bar, line, scatterplot combo of charts on a dashboard. But then I started thinking about the Fitbit dashboard. If I wanted to create a month view, what would I want?

So I started playing with calendars.

As you can see on the right, this is such a basic chart, but it still told me more than I knew from the scribbles up top.

Now this project was to try different things - So I didn't want to go *too* tried and tested, though this dataset was grating on me (an interesting thread which I read on Twitter earlier today talks about this, ending with this tweet from Chris Love here)

I wanted to add a bit more information within this, but also have different formats. Now, the end results could have easily been done in one label.. But I had a go trying a new way of formatting.

From here, I didn't think this could be a viz. It's a glorified formatted table! So I looked around the internet for some inspiration, but a lot of "sleep data" based visualisations are line charts, or dot plots. It's not an inspiring dataset it seems - Unless it's this one where a Dad visualised his daughter's sleeping patterns for the first 6 months of her life. I couldn't figure how to reviz it, but it could be something cool to come back to another time.

The next steps was simply plot my daily sleeping pattern day by day. Here I tried something different - instead of having an axis, or labels, I used reference lines to show the maximum and minimum, and customised the tooltips so it could be investigated later on.

The finishing touches were of course my favourite "metric cards" along the top, with some headers to spilt the dashboard up slightly.

Let me know what you think! (click to navigate)