Summary & Methods
Row
Days Tracked
14Locations Visited
6Avg. Mood
7.4Avg. Productivity
7Avg. Sleep (hrs)
6.9Column
About This Project
Data collection.
This dashboard is based on a personal activity log I built and tracked over a two-week period (July 28 - August 10, 2026), covering the places I regularly visited, how long I stayed, what I was doing, how much I slept the night before, and self-rated my mood and productivity (1-10 scale) for each activity block.
Why this approach.
I chose a manual daily log rather than an app export because none of my regular routines (studying, yoga, coffee shop visits, socializing) are captured by a single existing app, and building the log myself allowed me to define exactly the variables needed to answer my research questions, including a geographic component required for this assignment.
Tools used.
The data was cleaned and analyzed in R using tidyverse, leaflet for the geographic map, plotly for interactivity, and the whole dashboard was built with flexdashboard in R Markdown. I chose flexdashboard specifically because it does not require hosting on a server and it can be shared as a single, self-contained HTML file - an advantage that was noted in our final class session. For the interactive element, I used a plotly-wrapped scatterplot rather than a full Shiny app; Professor Cohen also noted that combining flexdashboard with Shiny is one way to get true reactivity, but that approach requires server-based hosting rather than a shareable static file, so plotly let me add genuine interactivity while keeping the dashboard portable. The log was stored as a CSV file, kept in the same project folder as this R Markdown file for reproducibility.
Assumptions.
- I assumed my self-rated mood and productivity scores (1-10) are consistent with myself over time, even though these are inherently subjective ratings without using an external validation method.
- I assumed the “Home” catch-all block period, which combine studying, downtime, and sleep recovery into a single row, are a reasonable simplification.
- I assumed my two weeks of tracked data are broadly representative of my typical routine.
- I assumed self-reported sleep hours are reasonably accurate, though not externally verified.
- I initially assumed more sleep would predict higher productivity, but my data showed a slight negative relationship instead — a reminder that self-tracked data can challenge intuitive assumptions rather than just confirm them.
Reading references applied.
Chart type choices follow Wexler’s principle of matching the visualization to the analytical question. Layout and clutter decisions follow Few’s (2008) guidance. The map uses a sequential color scheme since productivity has no natural midpoint (Juice Analytics, 2009). Tufte’s data-ink ratio principle guided the minimal, decoration-free chart style throughout.
My four research questions:
- Does my sleep the night before predict my mood or productivity?
- Where do I actually get my best work done, and does that match where I think I work best?
- Is there a day-of-week pattern to my mood, sleep, or productivity?
- How does time spent at a location relate to my mood there?
Where I Go
Column
Map of My Regular Locations, Colored by Average Productivity
About This Map
This map uses a sequential color scheme (Juice Analytics, 2009), appropriate since productivity has no natural midpoint. Circle size reflects total hours spent, color reflects average productivity.
Lake Mary (home to Craft & Common and SoFlo Hot Yoga) shows the darkest red marker, confirming the highest average productivity of any area I visit, which matches Question 2’s expectation. Sanford, where I live, sits in the middle of the scale — solidly productive but not my peak. Winter Park, where I socialize with my friend and run errands, shows the lowest average productivity, which makes sense given those visits are intentionally more relaxed and less work-focused.
Sleep & Weekly Patterns
Column
Sleep vs. Productivity
About This Plot
Position along an axis is the most perceptually accurate visual cue (Tufte), which is why a scatterplot with a trend line answers Question 1.
Interestingly, the trend line shows a slight negative relationship — more sleep is weakly associated with lower productivity in my data, the opposite of what I expected going in. Looking closer, this is likely driven by my Friday entries: several 7.5-hour sleep nights cluster around a productivity rating of only 5, pulling the trend downward, while some of my highest-productivity days (rating 9) happened on only 6 hours of sleep. This is a good example of why it’s worth checking assumptions against real data rather than trusting intuition alone — my two-week sample may simply be too small, or productivity that day may have been driven more by what I was doing (Craft & Common vs. an online class) than by sleep alone.
Column
Average Mood and Productivity by Day of Week
About This Plot
Grouping mood and productivity into one chart follows the “grouping structure” principle (Juice Analytics, 2009).
Friday stands out as a clear divergence point: it has my highest average mood (~9) but my lowest average productivity (~5) of the entire week — likely reflecting the “Exploratory Data Analysis class” days, which I enjoy but don’t associate with high output. Tuesday shows the opposite pattern, pairing high mood with my highest productivity of the week, aligning with my SoFlo Hot Yoga and Craft & Common routine. Overall, mood stays fairly high and stable across the week, while productivity swings more, suggesting my productivity is more sensitive to which specific activity I’m doing than my general day-to-day mood is.
Location & Duration
Column
Productivity by Location Type
About This Plot
A boxplot was chosen over a bar of averages because the question is about spread within each location type (Wexler).
Coffee Shop shows both the highest median productivity (close to 9) and almost no spread, confirming Craft & Common is my most reliable location for focused work. Gym follows closely behind, reinforcing the connection between exercise and productivity noted in the Summary tab. Home shows a single low point around 5, reflecting my “Studying/Downtime” catch-all — the widest range in this dataset comes from the mix of highly-productive class sessions and low-productivity downtime lumped into that one category. Friend’s House and Grocery Store sit lowest, which makes sense since those visits are for socializing and errands rather than focused work.
Column
Duration vs. Mood by Activity Type (Interactive)
About This Plot
This chart is the dashboard’s required interactive element, built with plotly so each point can be hovered to reveal its specific location, date, and duration.
The clearest pattern here is that my longest single block — a 10-hour “Studying/Downtime” day at home — has my lowest mood rating (5) in the entire dataset, while my shortest, most active blocks (Exercise/Socializing at ~1.5 hours, Socializing at ~2 hours) hit my highest mood ratings (9). This directly answers Question 4: more time spent doesn’t correlate with a better mood for me — if anything, my mood peaks during short, active, or social bursts, and drops during long, unstructured stretches at home.