Hi everyone 👋,
I’m a student currently working on my visualization project using the Lifestyle and Sleep Patterns
dataset from Kaggle (2025).
My goal is to explore how different lifestyle factors — such as age, physical activity, daily steps, and stress — affect sleep quality and duration.
I created four visualizations in Tableau Public:
1️⃣
Sleep Duration by Age Group (Bar Chart)
2️⃣
Physical Activity vs Quality of Sleep (Scatter Plot)
3️⃣
Daily Steps vs Quality of Sleep (Scatter Plot)
4️⃣
Stress Level vs Sleep Duration (Box Plot)
I tried to keep the design simple and clean — white background, light-blue color theme, and consistent labeling — to highlight the story rather than decoration.
However, since this is for an academic project, I’d love to get your advice:
- Do you think the color palette and layout are clear and effective?
- Are the relationships between variables easy to interpret from these charts?
- Any suggestions on how I could improve the storytelling flow or add interaction (filters, highlights, etc.)?
Any feedback, comments, or design tips are welcome! 💬
Thank you so much for taking the time to look at my work.
#Tableau Cloud
Things that jump out for me...
You have a blue and orange legend against the top left chart, but only blue bars. Be careful of obscuring your own data - if the bar chart is for the whole population, then you shouldn't have it as the same colour as one of your categories. You're obscuring data in the bottom right chart, where it appears that no males have a stress level of 4, or 8, which probably isn't the case. If you want to highlight the differenced between the genders then a side-by-side box-plot might give you a better illustration of the gender variance. If the stress measure isn't continuous (am I allowed a stress level of 6.328?), then strictly speaking a scatter plot isn't statistically valid. If people are asked to subjectively rate their stress, then a value of 7 is not a definitive point, in a similar way to a woman describing child-birth as a 9/10 for pain - a man describing stubbing his toe as a 10 isn't really a useful comparison. So, I would lean away from a scatter plot for that particular measure. One pitfall with a scatter plot in the situation you have is that a overlap obscures the density. Psychologically, I feel like there are very few people who have a stress level of 8, but there could be hundreds that all just have the same sleep duration. You could consider adding histograms to the top and right of your chart, if there are high densities that you think should be highlighted. This kind of thing - since the scatter plot values are relatively discrete, it seems like there are about 100 or so data points. The histogram makes it clear that the circles at the right of the chart represent far fewer data points than the circles in the middle:
That chart is also the only one where you haven't deselected "Show Zero" on the y-axis, so compared to the others it looks like all of the sleep durations are "high", where in fact they're the same, you've just started the other charts from 4 instead of 0. Where you've plotted lines of best fit, it would be helpful to users to have the r-squared value, because your line of best fit might explain 2% of the variation in quality of sleep, or it might explain 98%. Making claims about increasing steps helping with sleep can't simply be based on a line having a positive gradient.
