Business Area
Wearable technology, wellness analytics and user engagement
Google Data Analytics Capstone Project
Bellabeat Product Usage Analysis
Analysing Fitbit activity and sleep behaviour to identify user patterns and support data-driven wellness recommendations
Python • Pandas • NumPy • Matplotlib • Tableau • Data Cleaning • Feature Engineering • Correlation Analysis • Exploratory Data Analysis
This case study analyses Fitbit activity and sleep records collected between March and May 2016. It explores weekday behaviour, sedentary time, calorie expenditure, exercise intensity and sleep patterns before translating the findings into actionable recommendations for Bellabeat.
Project Overview
Clean Activity Data
1,373 unique daily activity records from 35 Fitbit users
Sleep Analysis Data
410 matched user-day activity and sleep records
Main Tools
Python, Pandas, NumPy, Matplotlib and Tableau
Business Problem
Bellabeat develops wellness products designed to help women monitor and improve their health. To strengthen its marketing strategy, the company needs a clearer understanding of how consumers use wearable fitness devices and which behavioural patterns may support more personalised engagement.
This project examines daily steps, calories burned, activity intensity, sedentary time and sleep behaviour to identify patterns that Bellabeat can use when designing reminders, activity goals, wellness campaigns and sleep-coaching features.
Key Business Question
What activity and sleep patterns can be identified from Fitbit data, and how can Bellabeat use those insights to improve customer engagement and support healthier lifestyles?
Dataset
The project uses Fitbit Fitness Tracker data collected during two reporting periods: March 12–April 11, 2016 and April 12–May 12, 2016. The activity datasets were combined, inspected and cleaned before being joined with daily sleep records.
Activity Metrics
Total steps, distance, calories, active minutes and sedentary minutes
Sleep Metrics
Total minutes asleep, total time in bed and sleep records
Time Variables
Activity date, reporting date and derived day-of-week feature
Analysis Type
Descriptive, exploratory, behavioural and correlation analysis
Analytical Process
The project follows the six phases of the Google Data Analytics process.
Ask
Defined the business task and analytical questions.
→
Prepare
Loaded activity and sleep datasets.
→
Process
Resolved overlaps, duplicates, dates and quality issues.
→
Analyze
Examined activity, calorie and sleep relationships.
→
Share
Created Python charts and a Tableau dashboard.
→
Act
Developed recommendations for engagement and wellness.
Behind the Analysis
Open each step to explore the Python workflow and analytical reasoning.
STEP 01Import LibrariesPandas, NumPy and Matplotlib were prepared for analysis.
Importing the Required Python Libraries
analysis.py
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings("ignore")
STEP 02Load Initial Activity DataThe first daily activity dataset was loaded into a Pandas DataFrame.
Loading the Initial Activity Dataset
STEP 03Explore Dataset StructureShape, sample records, data types, columns and missing values were reviewed.
Understanding the Dataset
explore_data.py
df.shape
df.head()
df.info()
df.columns
df.isnull().sum()Dataset Shape
Dataset Preview
Dataset Information
Column Names
Missing Values
STEP 04Combine, Clean and ValidateTwo activity datasets were combined and overlapping user-date records were resolved.
Preparing the Combined Activity Dataset
Load Both Periods
Daily activity records from both collection periods were loaded separately.
Concatenate
Matching structures allowed the two files to be combined.
Resolve Overlaps
Overlapping April 12 records were investigated and cleaned.
Validate Quality
The final activity dataset contained 1,373 records with no missing values or exact duplicates.
Load Both Datasets
Verify Dataset Shapes
Concatenate Datasets
Prepare Date Column
Verify Cleaned Shape
Duplicate Validation
Final Dataset Information
Data Quality Report
STEP 05Exploratory Data AnalysisActivity, calorie, sleep and weekday behaviour were explored.
Exploring Fitbit User Behaviour
Summary Statistics
Observation
Users averaged 7,377 daily steps and approximately 1,001 sedentary minutes.
Business Insight
Bellabeat can focus on reducing sedentary time and increasing daily movement.
Average Daily Steps by Weekday
Observation
Saturday recorded the highest activity, while Sunday recorded the lowest.
Business Insight
Sunday reminders and targeted challenges may help maintain weekend activity.
Daily Steps vs Calories Burned
Observation
Daily steps and calories showed a moderate positive correlation of 0.58.
Business Insight
Step goals are useful, but activity intensity should also be promoted.
Average Activity Minutes by Level
Observation
Sedentary time dominated the daily activity profile.
Business Insight
Movement reminders and gradual activity targets can support healthier routines.
Sleep Duration vs Time in Bed
Observation
Sleep duration and time in bed had a very strong positive correlation of 0.93.
Business Insight
Bellabeat can provide sleep-efficiency insights and personalised sleep goals.
Very Active Minutes vs Calories Burned
Observation
Very active minutes and calories burned showed a strong positive relationship.
Business Insight
Personalised vigorous-activity goals may improve fitness and engagement.
STEP 06Build Tableau DashboardThe Python findings were translated into a business-friendly dashboard.
Communicating Results Through Tableau
The final dashboard combines weekday activity, calorie relationships, activity intensity and sleep behaviour.
Visualizations
Six focused analyses explain activity, intensity, calories and sleep behaviour.
Summary Statistics
Establishes baseline activity, calorie and sedentary-time benchmarks.
Weekday Activity
Compares average daily steps from Monday through Sunday.
Steps and Calories
Measures the relationship between movement and calorie expenditure.
Activity Intensity
Compares active and sedentary minutes.
Sleep Behaviour
Examines sleep duration in relation to total time spent in bed.
Vigorous Activity
Evaluates how very active minutes relate to calories burned.
Tableau Dashboard
The dashboard communicates the major findings in a concise format suitable for decision-makers.
Bellabeat Fitness Activity Dashboard
Key Findings
The analysis revealed clear patterns in weekday activity, sedentary time, exercise intensity, calories and sleep.
Saturday was the most active day.
Sunday recorded the lowest average step count.
Users averaged approximately 7,377 daily steps.
The average indicates moderate activity with room for greater consistency.
Sedentary time dominated the daily routine.
Users averaged approximately 1,001 sedentary minutes per recorded day.
Steps and calories were moderately related.
The Pearson correlation coefficient was 0.58.
Sleep duration closely followed time in bed.
A correlation of 0.93 indicated a very strong positive relationship.
Vigorous activity increased calorie expenditure.
Very active minutes showed a strong positive relationship with calories burned.
Business Recommendations
The findings support personalised strategies focused on consistency, movement, exercise intensity and sleep quality.
Encourage Consistent Daily Activity
Use Sunday reminders, weekend challenges and personalised step goals.
Reduce Sedentary Behaviour
Introduce standing alerts, movement reminders and walking challenges.
Promote Higher-Intensity Exercise
Recommend short vigorous workouts and weekly exercise goals.
Improve Sleep Coaching
Provide bedtime reminders, sleep-efficiency feedback and personalised sleep goals.
Deliver Personalised Health Insights
Combine activity and sleep data to generate relevant coaching and progress tracking.
Executive Summary
This project analysed Fitbit activity and sleep data from 35 users between March and May 2016. Two activity datasets were combined and cleaned, resulting in 1,373 unique daily activity records. A separate sleep dataset was prepared and matched to activity data, producing 410 unique user-day records.
The analysis found that Saturday was the most active day and Sunday the least active. Users averaged approximately 7,377 steps per day but spent a substantial proportion of their recorded time being sedentary. Daily steps showed a moderate positive relationship with calories burned, while very active minutes showed a stronger relationship. Sleep duration and time in bed were very strongly related.
The findings support targeted activity reminders, sedentary-time reduction, personalised exercise goals and sleep-coaching features.
Skills Demonstrated
PythonPandasNumPyMatplotlibTableauData CleaningData ValidationFeature EngineeringExploratory Data AnalysisCorrelation AnalysisBehavioural AnalyticsData VisualizationBusiness InsightsData Storytelling
Project Resources
Replace the placeholder links below with your published Kaggle and Tableau URLs.