bellabeat_case_study

Bellabeat wanted to understand smart-device wellness behaviour and identify opportunities to strengthen marketing strategy.

• SQL (Google BigQuery)
• Python (Pandas)
• Tableau

1. Data cleaning and preparation
2. SQL analysis
3. Python correlation analysis
4. Tableau visualizations
5. Business recommendations

• Low Overall Activity: Users average ~7,377 steps/day, with >53% of days classified as sedentary or low active. Only 31% hit the 10,000-step benchmark.
• High Sedentary Time: Daily logs show an extreme ~1,001 sedentary minutes compared to just ~20 minutes of high activity.
• Sleep Impact: Increased step counts do not extend sleep duration, but higher sedentary time significantly reduces sleep ($r \approx -0.60$).
• Calorie Burn: Energy expenditure increases moderately alongside step counts ($r \approx 0.58$).
• Peak Active Windows: Daily movement and intensity peak consistently in the afternoon and evening (12:00, 14:00, and 17:00–19:00).

Tableau Desktop was used to create visualizations exploring activity levels, sleep behaviour, and daily wellness patterns.

The dashboard includes:

• Step Goal Distribution
• Activity Level Distribution
• Average Sleep by Activity Level
• Daily Wellness Rhythm
• Daily Activity Heatmap

A screenshot of the dashboard is included within this repository along with the Tableau workbook file.

1. Promote Gradual Movement Goals and Habit Formation
2. Target Sedentary Behaviour Through Smart Wellness Prompts
3. Market Wellness as a Balance Between Movement and Recovery
4. Use Peak Activity Hours for Timely Engagement
5. Leverage Personalization Through Behaviour Profiles

This repository contains all materials developed for the Bellabeat case study completed as part of the Google Data Analytics Certificate.

Contains the final written case study report, including the business task, data cleaning process, analysis, visualizations, findings, and recommendations.

Files:

• Bellabeat_Report.pdf

Contains SQL queries written and executed in Google BigQuery for data cleaning, transformation, merging, and exploratory analysis.

Files:

Contains SQL scripts and query results completed in Google BigQuery for data cleaning, table preparation, and exploratory analysis.

The SQL work is organized into separate folders based on each stage of the project:

Folders:

• cleaning

Contains data cleaning and preparation queries, including duplicate checks, timestamp conversion, column standardization, and data validation.

• table setups

Contains table creation, merging, and preparation scripts used to build cleaned and analysis-ready datasets.

• 01. User Patterns Analysis

Contains descriptive statistics and user activity pattern analysis queries.

• 02. Activity Level Segmentation Analysis

Contains queries used to classify and analyze activity levels using Tudor-Locke and Bassett step categories.

• 03. Activity vs Sleep Analysis

Contains SQL queries examining relationships between movement and sleep behaviour.

• 04. Hourly Activity, Intensity, and Energy Expenditure Patterns Analysis

Contains hourly movement, intensity, and MET analyses used to identify behavioural peak periods.

• Query Results

Contains exported query outputs and analysis tables generated during the project.

Contains the Jupyter notebook and exported analysis used for data validation, correlation analysis, and behavioural investigation using pandas.

Files:

• Bellabeat_Analysis.ipynb
• Bellabeat_Analysis.pdf

Contains Tableau dashboard materials and links used to visualize findings related to activity, sleep, and wellness behaviour.

Files:

• Tableau workbook (.twb)

Contains screenshots of dashboard visualizations and supporting graphics used within the project and GitHub README.

Files include:

• Step Goal Distribution
• Activity Level Distribution
• Sleep by Activity Level
• Daily Wellness Rhythm
• Activity Heatmap
• Dashboard overviews

The original Fitbit dataset used in this project was sourced from Kaggle (Mobius) under the CC0 Public Domain license.

To keep the repository organized and lightweight, raw datasets are not included.

Source:

• Fitbit Fitness Tracker Data (Kaggle/Mobius)