This repository contains my capstone project for the Google Data Analytics Professional Certificate. The project is a real-world case study where I act as a junior data analyst for Bellabeat, a high-tech company that manufactures health-focused smart products for women.
The central business task was to analyze user data from a competitive product (Fitbit) to uncover insights into consumer behavior. The final goal is to deliver data-driven recommendations that can guide Bellabeat's marketing strategy and product development.
This project rigorously follows the six steps of the data analysis process as taught by Google: Ask, Prepare, Process, Analyze, Share, and Act.
- Business Objective: Identify trends in smart device usage to inform Bellabeat's growth opportunities.
- Data Vetting: The initial phase involved a critical evaluation of the provided dataset (Fitbit Fitness Tracker Data from Kaggle). I identified key limitations, including a small sample size (30 users), outdated information (from 2016), and a lack of demographic data, which are crucial considerations for a real-world business scenario.
This phase was focused on transforming 18 raw, disorganized CSV files into a single, unified, and clean dataset ready for analysis.
- Data Cleaning in R: Used the Tidyverse suite in R to process the majority of the datasets. This involved:
- Standardizing date and time formats across all files.
- Merging multiple tables (e.g.,
dailyActivity,sleepDay,weightLogInfo) into a cohesive master table. - Checking for and handling inconsistencies.
- Handling Big Data with SQL: The
heartrate_secondsdataset was too large for efficient processing in memory with R. I demonstrated adaptability by:- Uploading the data to Google BigQuery.
- Using SQL queries to clean the data and aggregate the second-level heart rate data into hourly averages.
- Importing the processed, smaller dataset back into R to merge with the main table.
With the clean, unified dataset, I performed a deep exploratory data analysis to uncover user habits.
- Key Findings:
- A strong positive correlation was found between total steps and calories burned, confirming that daily activity is a key driver of energy expenditure.
- Analysis of activity intensity revealed that users spend the vast majority of their time in sedentary or lightly active states, with very few minutes of high-intensity activity.
- Sleep analysis showed most users get within the recommended 7-9 hours of sleep, but there is a negative correlation between sedentary time and sleep quality.
- Visualizations: The analysis is supported by a variety of visualizations created with
ggplot2in R, including scatter plots, bar charts, and time-series plots to make the findings clear and actionable.
The project culminates in a set of strategic recommendations for Bellabeat, directly derived from the data insights:
- Recommendation 1 (Marketing): Launch targeted campaigns encouraging users to convert light activity into moderate activity, highlighting the significant calorie-burn benefits.
- Recommendation 2 (App Features): Implement a "sedentary alert" feature in the Bellabeat app that nudges users to move after prolonged periods of inactivity, explaining the positive impact on sleep.
- Recommendation 3 (Product): Develop personalized daily and weekly summary reports that focus on key achievements (like total steps and sleep quality) to keep users engaged and motivated.
- Languages: R, SQL
- R Packages:
Tidyverse(includingdplyr,ggplot2),janitor,skimr - Database: Google BigQuery
- Environment: RStudio
- Clone the repository.
- The repository contains the R Markdown file with the complete, documented code.
- The raw data can be downloaded from this Kaggle link.
- Follow the steps in the notebook to reproduce the entire analysis from data cleaning to final recommendations.
Antonio Barrera Mora
- LinkedIn: https://www.linkedin.com/in/anbamo/
- GitHub: @Kamaranis