Used real Fitbit activity and sleep data to understand how wellness habits translate into smart device engagement. The analysis surfaces patterns Bellabeat can use to shape product features and targeted marketing for health-conscious users.
Project overview
Explored how people actually use their wellness wearables across a typical month. By combining daily steps, calories burned, active minutes, and sleep data from Fitbit, built a picture of when users move, rest, and engage with their devices.
Key deliverables
Phase 1
Ask
Define how device usage and wellness behavior can guide Bellabeat's engagement strategy.
Phase 2
Prepare
Import Fitbit activity, sleep, and calories data for 30 users across a month.
Phase 3
Process
Clean records, standardize dates, and engineer daily wellness metrics.
Phase 4
Analyze
Explore activity and sleep trends across users and days of the week.
Phase 5
Share & act
Translate findings into engagement and marketing recommendations for Bellabeat.
The analysis uses a public Fitbit dataset with about 30 users observed over 30 days. It includes daily steps, distances, calories, intensity minutes, and separate tables for sleep and weight. The guiding question is how these patterns can help Bellabeat increase activation, retention, and daily engagement with its own smart devices.
Bellabeat wants to know when its users are most active, how consistently they sleep, and how these behaviors line up with device usage. By treating Fitbit users as a proxy for Bellabeat's target audience, this analysis prototypes the kinds of metrics and visuals that would power Bellabeat's wellness app and marketing campaigns.
Merged Fitbit dataset preview with daily activity metrics for each user.
To keep the project focused, a subset of fields was selected that best capture daily wellness: steps, total distance, calories burned, active minutes at different intensities, and sleep duration. These map to KPIs Bellabeat could track in its own product.
R environment prepared with tidyverse, lubridate, janitor, and ggplot2 for the analysis.
Activity and sleep tables did not share a perfect key out of the box. Date formats were standardized, duplicate entries removed, and tables joined at the user and date level. From there, features were engineered that describe a typical day for each user.
The merged activity data needed basic hygiene before it could be trusted. Records with missing or zero steps were filtered out, date fields converted to proper date types, and column names aligned across tables.
Cleaned dataset with engineered metrics such as total active minutes and weekday labels.
After joining activity and sleep tables on user + date, features were engineered that Bellabeat could track as KPIs: total active minutes, sleep minutes, and day of week. These become the foundation for segmenting users and spotting opportunities.
Sample R — merge and engineer features
library(tidyverse)
library(lubridate)
library(janitor)
daily_activity <- read_csv("dailyActivity_merged.csv") %>%
clean_names()
merged <- daily_activity %>%
inner_join(daily_sleep, by = c("id", "activity_date")) %>%
mutate(
day_of_week = wday(activity_date, label = TRUE),
active_minutes_total = very_active_minutes +
fairly_active_minutes +
lightly_active_minutes
)
With a clean daily view of each user, examination focused on how steps, active minutes, calories, and sleep changed by day of week. This highlights when users are most engaged with their wellness routine and where there may be drop-off.
Average steps and calories peak midweek and dip on weekends. This suggests that structured routines (work, commute, gym) drive more consistent movement than unstructured weekend time.
Average daily steps by weekday — weekdays outpace weekends, especially midweek.
A scatter plot of total active minutes versus calories burned shows a clear positive relationship: more movement reliably translates into higher energy expenditure.
Strong positive relationship between total active minutes and calories burned.
Most users sleep between six and seven and a half hours per night, which is below the recommended eight hours. This creates an opportunity for Bellabeat to position sleep insights as a core value add.
Histogram of sleep duration with the recommended 8 hours highlighted as a target.
Once the core metrics were engineered, R and ggplot2 were used to create visuals and summary tables that make the data easy to read for stakeholders — even if they don't work with R themselves.
A line chart of average calories by weekday reveals midweek peaks and weekend drop-offs. This directly supports the idea of targeting programs when users are already in motion.
Average calories burned by weekday — midweek peak and weekend drop are clear visual cues.
A simple summary table helps translate visuals into numeric benchmarks that Bellabeat could use in dashboards or user segmentation logic.
For each day of the week, average steps, calories, and sleep duration were calculated. This makes it easy to compare performance at a glance and choose reasonable goal ranges.
Summary of average steps, calories, and sleep minutes by weekday.
This R code groups the merged dataset by weekday and computes average steps, calories, active minutes, and sleep minutes — the core KPIs that power the visuals above.
day_summary <- merged %>%
group_by(day_of_week) %>%
summarise(
avg_steps = mean(total_steps, na.rm = TRUE),
avg_calories = mean(calories, na.rm = TRUE),
avg_active_minutes = mean(active_minutes_total),
avg_sleep_minutes = mean(total_minutes_asleep),
.groups = "drop"
) %>%
arrange(day_of_week)
Once patterns were clear, they were translated into concrete actions Bellabeat could take. The focus is on using data to time nudges, adjust messaging, and design challenges that align with how users already live their lives.
The calories-by-weekday view directly supports timing campaigns around midweek activity peaks.
Composite dashboard bringing together activity, sleep, and calorie insights for Bellabeat stakeholders.
Analysis of Fitbit user behavior reveals clear patterns in daily wellness habits that can inform Bellabeat's product strategy and user engagement approach.
Most users fall short of eight hours of sleep even on low activity days.
Weekday activity is more structured, making it easier to anchor challenges and streaks.
Weekend engagement dips suggest a need for lighter, lifestyle-oriented nudges.
Users with more consistent sleep and activity patterns show more stable device usage.
Objective: Leverage structured weekday routines to build lasting activity habits.
Objective: Position sleep tracking as a core value proposition for wellness.
Objective: Maintain engagement during unstructured weekend periods.
This Bellabeat smart device analysis connects raw Fitbit data to concrete engagement opportunities. By cleaning and joining activity and sleep records, engineering daily wellness signals, and studying behavior by day of week, the project shows when users are most receptive to coaching and how Bellabeat can time its features, challenges, and messaging to match real user habits.