User Experience Stack Exchange is a question and answer site for user experience researchers and experts. It only takes a minute to sign up.
Sign up to join this community
I need to present data about defects that "I" created in the last 7 days, and want to show 4 different aspects:
Here's what I've come up with so far:
Yes, I know this is hideous; it's just a prototype and it would actually have a y-axis (count) scale and a color legend (where color = severity).
I don't find this visualization particularly appealing (beyond even the ugly-prototype-y-ness of it) and I wonder if there's a better format.
Usually charts like this show the most recent data on the right, not the left, so I'd reverse the x axis.
Stacking by severity is a good way to show both count and severity in a single column.
However, you don't have comparable things next to each other. By combining bugs, violations and smells for the same date together, you're forcing your eye to compare them with each other, instead of (for example) allowing bugs to be compared across time.
For this reason, I'd suggest splitting into three stacked charts with a common x-axis - here's a mockup I threw together in Excel:
Can I make a book recommendation? Edward Tufte's 'The Visual Display of Quantitative Information'. You'll never look back.
From a usability point of view - once you've sorted out your colours and simplified your lines - the key here is to allow users to manipulate the data in many different ways. For example, is it important to know on which days you get the most errors? What are the trends for severity? Can you correlate the data against your release cycle over time etc etc.
I had fun with your problem (see ppt progression below). I'm not trying to present the best graph type for you, as you'll have to experiment a bit. But you can see how colour and layout can help present information in a more digestible way. I'd stay away from the stacked line graphs personally in favour of separating the graphs out and aligning them horizontally or vertically.
Hideous prototypes are better at this stage anyway, keeps the focus on the concept and not the details.
My first thought on this one is to have three stacked line graphs, one for each type of defect with the stacks in each representing the severity groupings (low/medium/high) all sharing a single time axis.
You will have to include a legend somewhere for severity levels (would recommend a more red to blue scale instead of red to green to help color blind users).
You will also have to decide if you need to compare counts of bugs, violations, and code smells to each other. If so they need to share the same range on the y-axis. If not you can make the scales different and avoiding having one type with far more defects than the others from squashing the detail in the others.
The shortcoming of this design is that it is not optimized for comparing the same severities across defect types except for the very first severity (the red one in my mockup). This is because the yellow and blue stacked lines do not have a stable baseline and humans have a perceptual bias toward seeing the width of the band as the shortest line between the parallel edges instead of the actual vertical drop.
You can overcome it, but your at-a-glance interpretation is a bit off because of it.
I think the answer would really depend on what you want to focus on with the data representation. I assume that you have combined the three different types of defect types because you want to compare them on a day-to-day basis, otherwise if you want to compare the same type of defect over a time period then you are better off using separate charts so as to not get lost in the sea of information. You are also introducing severity using colour, which would make it hard to compare the number of each severity level for each defect type. You'll need to construct the visualization based on what you want to focus on with the comparisons, or if you are showing trends or patterns then you'll need to focus on how best to make those insights clear.
If you insist on trying to capture all the complexity, it might also be worth thinking about a table formatted to highlight areas that you need to focus on. I have provided an example below that put in red all the areas that simply applies a conditional formatting rule on excel, but this allows you to mark in red all the sections that have higher values compared to the same category. I haven't given much thought to colours and styling but hopefully you get the idea.
