Bellabeat Capstone_ Smart Device Usage Analysis.pdf
As the capstone project for the Google Data Analytics Certificate came to an end, I completed a case study focused on helping Bellabeat analyze fitness tracker data to identify marketing opportunities and better understand consumer wellness behaviors.
Using publicly available fitness data from other wellness companies, I cleaned, organized, and analyzed datasets to identify trends in user activity, sleep, calories burned, and overall fitness habits. I used RStudio to process and visualize the data, creating charts and graphs that highlighted relationships between key health metrics and user engagement patterns.
Throughout the project, I followed the full data analysis process, including asking business-focused questions, preparing and cleaning data, performing exploratory analysis, interpreting findings, and developing actionable recommendations. I then created and presented a detailed slideshow summarizing the project objectives, analytical process, visual findings, and strategic recommendations for how Bellabeat could improve future marketing efforts and customer engagement.
When researching a topic, I had to come up with the reason I am doing it. In this instance, I was challenged with coming up with marketing and engagement strategies for the company Bellabeats, after analyzing data from other fitness websites. I have a major interest in fitness, so I accepted the challenge.
I came up with a few research leading questions to help with my research, these were:
Are there patterns in activity across weekdays? ● How active are users overall? ● How do users spend their active minutes? ● How does activity impact calories burned? ● Are sleep patterns consistent across weekdays/weekends? ● How often do users meet step goals?
Using Microsoft Excel and RStudio, I began the analysis by downloading the Fitbit Fitness Tracker dataset from Kaggle. A critical part of the project involved cleaning the data to remove inconsistencies, correct formatting issues, and ensure the datasets were accurate and reliable for analysis. After cleaning the data, I merged multiple datasets using participant ID and date fields, which allowed activity, sleep, calorie, and wellness metrics to be connected for each individual user over time. This step was essential for identifying relationships between different fitness measurements and creating more meaningful insights from the combined data.
After preparing and merging the datasets, I answered the six key business questions by conducting exploratory data analysis in RStudio. Using ggplot2, I created visualizations that identified patterns, trends, and relationships in the fitness data. One area of analysis I focused on was whether users were consistently meeting the recommended 10,000 daily step goal. By writing R code to categorize users based on step performance and grouping the results, I created a bar graph that clearly visualized how often users met or failed to meet the goal. I found that 50% more days were completed without the step goal being achieved.
I also analyzed how users spent their time throughout the day and examined relationships between physical activity and calories burned. Using ggplot visualizations such as scatter plots and pie charts, I was able to identify activity trends, compare user behaviors, and better understand how movement levels impacted overall fitness measurements. In this example, I found that majority of peoples’ days are spent in lightly active minutes. I found that 85% of the minutes tracked were spent in lightly active minutes. These visualizations made the data easier to interpret and allowed me to communicate findings and recommendations more effectively within my final presentation.
By the end of this project, my analysis skills in RStudio had significantly improved, particularly in data cleaning, visualization, and interpreting trends through exploratory analysis. I also learned that meaningful analysis takes time, patience, and iteration, but the process itself is highly engaging and rewarding as patterns begin to emerge from the data.