A while ago, I started tracking my weight daily and having healthier habits, with the hope of losing some of it.I. The ProblemAfter a few days of tracking my weight with my Fitbit Aria, all the excitement got eroded with the limitations of their data visualizations. Fitbit, and as far as I'm concerned, many weight trackers display time-series linearly. There's nothing wrong with line charts, but sometimes, they hide information.For example, Fitbit's dashboard displays in a line-chart a week worth of weight/BMI data.
The problem with the image above, is that for someone that has been trying to lose weight, or even worse, someone that has problems trying to lose some weight, it might be a bit discouraging to see it, as data for a week might be missing the big picture; see the image below.
For these kind of users, the big picture is what matters more than the day to day.So I jumped to re-imagine the data display that would tackle the problem of weight-loss effectively.
The metrics were exactly the same that you'd see in fitbit's dashboard or any other website or app that allows you to track weight. Weight and BMI tied together with a date; after a week I had seven data points with two metrics and a some kilos off me.
Changing the display to show you all the "progress that matters" starts adding clutter to the chart. And every time the body decides to fight some weight back, makes your anxiety and panic levels go high.
It's evident that there should be something that you could do to this screen. The weight in the line-chart was going up and down from one day to another, but it is rather meaningless; However, on a weekly basis, data may uncover different insights. So let's break the classic line-chart time-series into many weekly series, so that I could see the progress week after week.The BMI color-scheme in the background adds some benchmark reference to the weight-values; the weight-value alone is almost useless keeping fitness in mind.
However, as weeks are added, the chart gets cluttered again but we might be able to get some important insight from it; the sense of progress.
Progress, at least for anyone dealing with weight problems, is measured in kilos (or equivalent) per unit of time. In the scenario depicted in the chart above, progress is visualized week by week. Since data granularity allowed it, calculations were performed to transform the data into weight-loss (in kilos) per day.
And then, since we agreed that the BMI reference is required, we decided to plot it in as well.
And the problem of data visualization just got a little bit more complicated, as the chart was difficult to understand, but it also yields some fantastic insights.First, we are able to determine the change in weight-loss ratio (kilos/day). Then, by taking into consideration the last 30 measurements (Central Limit Theorem) it is possible to find the mean and the standard deviation.
With these three values, a line that intersects the BMI curves or any other arbitrary weight-line (i.e. a Weight-goal weight line) can be traced; using two standard deviations, we get a good a good confidence interval (especially If people weighing themselves keep the same lifestyle that they've had for the last 30 days).
If we keep track of weight values daily, we would end up with a chart similar to the one below..
III. The ResultsAs I did this initially for myself, I had to test first that my hypothesis would be right, so I recorded some of my parameters one day and time-stamped them with twitter.My parameters were:
I wrote this tweet to time-stamp the forecast.
Maths+Stats predicted on 19-Jan (-11.7kg then) that I'd reach -16.1kg on 12-Feb +/-1d. I started (0kg) on 11-Nov. #weightloss— Edgar Anzaldúa (@edgarator) January 26, 2014
And this is the result on the date predicted.
@edgarator -15.8kg today 11-Feb, after predicting -16.1kg 3 weeks ago. Error 0.3kg (Not too bad)! #weightloss @fitbit — Edgar Anzaldúa (@edgarator) February 10, 2014
IV. The OutcomeThe whole process of measuring data, learning from it and designing an overly complicated dashboard gave me an epiphany. I started to clean up the mess.
Since we were talking of progress as the main focus for people trying to control their weight. The first thing that may come to your mind may be a progress bar, since progress bars convey straight off the bat the sense of... err... progress.However, weight behaves arbitrarily without us having much control over it. So there's the chance that instead of going right, towards completion, you might go backwards (to the left, towards the start day), which was nonsensical, or at least as a concept many not help much conveying the right story.
A better metaphor to tell the story may be a scale's gauge. The beauty of it is that the current value always sits in the middle, and just with a little tweak you could still convey the sense of the progress.
The image above, displays information for someone that in a given window of time.
Using the gauge alongside a filter allows you to see time-boxed progress in a snap-shot.
In the image above, you can see how the user has
V. Additional Design ConsiderationsWe all know that while weight tracking you might have three different kinds of users:
The gauge only adds the ability to play with radial motion, the way a physical scale does. Even if the gauge only has a limited value range, let's say current weight +/- 45 kilos, as opposed to having a full ranged fixed scale that goes from 0 to 160 kilos (as my scale at home does), the gauge still provides the sense of balance between positive and negative progress with the green and red bars. But of course, the value range could be easily adjusted.V. The App IdeaAfter a lot of fiddling a couple of friends and I started to having a go at building a website that would lead into this, but not being our main focus in life, and having difficulty to figure out an easy way of monetizing the application, we decided to put the idea on the drawer.But this is where we were at in concept design:
One last thing. One of the design issues that we faced was the fact that we wanted the users to log in their weight as often as possible so that we could predict the direction and magnitude of weigh-loss tomorrow. We thought, that given how easy is for people to give up on weight-loss when they face a negative outcome, if we tell them before hand that they are likely to put on weight tomorrow, that expectation could help dealing with the emotional and behavioral side-effects. The background (wall-paper) would change, depending on weather conditions and positive and negative weight variation. The idea behind this was to provide some eye candy, and as a way to provide a Variable Reward, which in words of Nir Eyal, author of "Hooked, How to Build Habit-Forming Products" reads like:
What distinguishes the Hook Model [...] is the [...] ability to create a craving. [...] The unsurprising response of your fridge light turning on when you open the door doesn’t drive you to keep opening it again and again. However, add some variability to the mix — say a different treat magically appears in your fridge every time you open it — and voila, intrigue is created.
And we thought it was fun to add some nice pics into the mix as well.
