Business Task and Problem Statement

Problem Statement:
Bellabeat seeks to understand how consumers use smart devices to make data-driven decisions that enhance the marketing strategy for its products. By analyzing trends in smart device usage, Bellabeat aims to identify opportunities to optimize its marketing strategy and strengthen its position in the global smart device market.

Business Task:
Analyze available smart device usage data to identify key trends and insights. Then, apply these insights to one Bellabeat product to provide actionable recommendations for enhancing Bellabeat’s marketing strategy.

Guiding Questions and Key Tasks

1. What is the problem you are trying to solve?

• Bellabeat aims to leverage data insights to understand how users interact with smart devices and identify growth opportunities.

2. How can your insights drive business decisions?

• Insights from the data can inform targeted marketing campaigns, feature improvements, and customer engagement strategies for Bellabeat's products.

Key Tasks:

• Identify trends in smart device usage, such as patterns in activity, sleep, stress management, or hydration tracking.
• Assess how these trends align with the behavior and preferences of Bellabeat’s target audience.
• Provide actionable recommendations for one Bellabeat product based on the analysis.

Key Stakeholders:

• Urška Sršen: Bellabeat co-founder and Chief Creative Officer.
• Sando Mur: Co-founder and mathematician, providing technical and strategic support.
• Bellabeat Marketing Analytics Team: Team responsible for collecting, analyzing, and leveraging consumer data.
• Bellabeat Customers: Existing and potential customers of Bellabeat products.

Deliverable

A clearly articulated business task statement:
“Analyze smart device usage data to uncover consumer trends and provide actionable recommendations for enhancing the marketing strategy of Bellabeat's products, focusing on one specific product to maximize its appeal and market reach.”

Data Preparation for Fitbit Fitness Tracker Data

Guiding Questions

1. Where is your data stored?
   The data is stored on Kaggle, accessible via the public dataset: FitBit Fitness Tracker Data. It has been provided under the CC0: Public Domain license, ensuring no copyright restrictions.
2. How is the data organized? Is it in long or wide format?
   The dataset is organized in a wide format. Each row corresponds to a user's daily record or minute-level data for specific variables such as activity levels, steps, heart rate, and sleep monitoring. Columns represent variables (e.g., TotalSteps, Calories, SleepMinutes).
3. Are there issues with bias or credibility in this data? Does your data ROCCC?

• Reliable: The dataset has minute-level details, which indicate a high degree of granularity. However, it is based on data from only 30 users, limiting representativeness.
• Original: Data was directly sourced from Fitbit trackers, indicating minimal external tampering.
• Comprehensive: Data spans several activity dimensions (daily steps, sleep, and heart rate) but does not cover diverse demographics.
• Current: The dataset is relatively dated and may not reflect current trends in fitness tracking.
• Cited: Proper consent was obtained from participants.

4. This dataset may exhibit biases due to its small sample size and lack of diversity. Additional data from a broader population can address these limitations.
5. How are you addressing licensing, privacy, security, and accessibility?

• Licensing: The dataset is available under CC0: Public Domain, allowing unrestricted use.
• Privacy: Personal identifiers are excluded, ensuring anonymity.
• Security: Data is securely downloaded and stored in a controlled environment.
• Accessibility: Data is in CSV format, which is universally accessible for analysis.

6. How did you verify the data’s integrity?
   Data integrity is verified by:

• Checking for missing values and inconsistencies.
• Ensuring logical data ranges (e.g., steps cannot be negative, heart rates within typical human limits).
• Inspecting dataset size to ensure no corruption during download.

7. How does it help you answer your question?
   The data provides detailed insights into daily habits (physical activity, sleep, and heart rate), allowing exploration of correlations and patterns in user behavior. However, its narrow demographic scope suggests complementing it with other datasets for generalizability.
8. Are there any problems with the data?

• Small sample size limits statistical significance.
• Demographic diversity (age, gender, geographic location) is unknown.
• Potential inaccuracies due to Fitbit device measurement errors.

Key Tasks

1. Download data and store it appropriately:
   Data is downloaded from Kaggle and stored in a dedicated project folder with proper versioning for safety and reproducibility.
2. Identify how it’s organized:

• Observed wide-format data with columns representing variables and rows representing users/daily records.
• File inspection reveals several CSV files for different time scales (daily, minute).

3. Sort and filter the data:

• Filtered out incomplete or anomalous entries (e.g., days with zero activity).
• Sorted data chronologically for trend analysis.

4. Determine the credibility of the data:

• Verified consent process and checked Kaggle's dataset documentation.
• Evaluated data origin (directly from Fitbit devices).

Deliverable

Description of Data Sources Used:

1. FitBit Fitness Tracker Data

• Source: Kaggle
• Content: Physical activity, sleep monitoring, heart rate, and daily steps data for 30 Fitbit users.
• Licensing: CC0: Public Domain
• Limitations: Small sample size, unknown demographic diversity.

2. Complementary Data (Optional for addressing limitations):

• Additional datasets from open data repositories (e.g., WHO, national health surveys) can provide broader demographic insights for comparison and validation.

Spreadsheets

SQL

R

Impofiing data from spreadsheets: This video from Course 3 will guide you through the steps you will follow to impofi data into your spreadsheet. This is useful if you want to pefiorm your cleaning and analysis with spreadsheets.

Uploading a CSV file in BigQuery: These step-by-step instructions will guide you through the process of uploading your CSV file into BigQuery so that you can stafi working with your data in SQL.

Data import basics: This reading from Course 7 will review importing data into R so that you can begin cleaning and analyzing it. If you are planning to use R for your case study, this is a useful starting point.

Data cleaning features in spreadsheets: This video from Course 4 outlines basic data cleaning features in spreadsheets; this is a great refresher if you need a review.

Cleaning string variables using SQL: This video from Course 4 covers some key cleaning techniques for string data in SQL.

Cleaning up with the basics: This video from Course 7 will guide you through some basic R cleaning functions that you will need to process your data for analysis.

Even more data-cleaning techniques: This video from Course 4 covers even more techniques you can use to clean your data and prepare it for analysis.

Advanced data-cleaning functions pafi 1 and pafi 2: These videos cover more advanced cleaning functions that are a great refresher as you begin to work more closely.

Transforming data: This video from Course 7 covers transforming data in R so that it is organized and formatted for easy analysis.

Process: Data Cleaning and Preparation for Analysis

1. What tools are you choosing and why?

• R Programming (tidyverse):
  The tidyverse package is chosen due to its comprehensive suite of tools for data manipulation, visualization, and analysis. It simplifies cleaning, transforming, and summarizing large datasets efficiently.

2. Have you ensured your data’s integrity?

• Data integrity was maintained by:

• Inspecting for missing values: Both datasets were checked, and no missing values were found (sum(is.na()) results were 0 for both datasets).
• Removing duplicate rows: Duplicate rows were identified and removed using the distinct() function.
• Ensuring consistent date formatting: Dates were reformatted uniformly using as.Date().
• Aligning data types across columns: str() was used to confirm correct column types.

3. What steps have you taken to ensure that your data is clean?

• Steps performed include:

• Removing duplicates from daily_activity and sleep_day datasets.
• Converting dates into a consistent format (%m/%d/%Y) to align across datasets.
• Filtering out invalid or extreme outlier values in numerical columns such as TotalSteps and TotalMinutesAsleep.
• Dropping unnecessary columns (e.g., SleepDay after merging datasets).

4. How can you verify that your data is clean and ready to analyze?

• Verification measures:

• Summary statistics using summary(): Ensured no extreme anomalies and calculated central tendencies (mean, median).
• Visual inspections: Scatterplots were created to identify trends or anomalies, such as:

• Relationship between TotalSteps and SedentaryMinutes.
• Relationship between TotalMinutesAsleep and TotalTimeInBed.

• Dataset merging: Merged datasets by Id and confirmed alignment of rows and columns.

5. Have you documented your cleaning process so you can review and share those results?

• The entire process was documented in an R script, covering:

• Data import steps (read.csv()).
• Cleaning operations (e.g., distinct(), mutate() for derived metrics).
• Transformation logic, including derived metrics like:

• SleepEfficiency = TotalMinutesAsleep / TotalTimeInBed
• ActivityToSedentaryRatio = TotalSteps / SedentaryMinutes

• Export of cleaned datasets using write.csv().

Key Tasks

1. Check the data for errors:

• Missing values, duplicates, and data type mismatches were identified and resolved.

2. Choose tools:

• tidyverse in R for robust cleaning and visualization.

3. Transform the data:

• Calculated derived metrics (SleepEfficiency, ActivityToSedentaryRatio).
• Summarized metrics (e.g., average steps and sleep efficiency by user).

4. Document the cleaning process:

• All steps are logged in the below document for reproducibility and sharing.

Data Cleaning and Manipulation Documentation

Tools Used

• R Programming Language

• Libraries: tidyverse for data manipulation and visualization.

Steps Taken to Clean and Prepare the Data

1. Data Import

• Loaded daily_activity and sleep_day datasets using read.csv().

2. Initial Exploration

• Explored datasets to understand structure and contents:

• Used head() to preview the first few rows.
• Listed column names with colnames().
• Verified data types using str().

3. Checking for Missing Data

• Verified the presence of missing values in both datasets using sum(is.na()).

• Result: No missing values found.

4. Removing Duplicates

• Checked for duplicate rows using distinct().
• Removed duplicate rows from both datasets:

• daily_activity <- daily_activity %>% distinct()
• sleep_day <- sleep_day %>% distinct()

5. Ensuring Consistent Date Formatting

• Reformatted dates to ensure consistency:

• daily_activity$ActivityDate <- as.Date(daily_activity$ActivityDate, format = "%m/%d/%Y")
• sleep_day$ActivityDate <- as.Date(sleep_day$SleepDay, format = "%m/%d/%Y")

6. Alignment of Column Data Types

• Verified column data types and ensured alignment using str().

7. Filtering Invalid or Extreme Values

• Removed rows with invalid or extreme outlier values:

• Ensured all values in TotalSteps, TotalMinutesAsleep, and SedentaryMinutes were ≥ 0.
• Used filter() to retain only valid data.

8. Merging Datasets

• Merged daily_activity and sleep_day datasets on Id and ActivityDate using merge():

• combined_data <- merge(sleep_day, daily_activity, by="Id")

9. Removing Redundant Columns

• Dropped unnecessary columns, such as SleepDay, after merging datasets:

• combined_data <- combined_data %>% select(-c(SleepDay))

10. Derived Metrics

• Created new metrics to enhance analysis:

• Sleep Efficiency: SleepEfficiency = TotalMinutesAsleep / TotalTimeInBed
• Activity-to-Sedentary Ratio: ActivityToSedentaryRatio = TotalSteps / SedentaryMinutes

11. Summarization by User

• Aggregated data to calculate average values per user:

• user_summary <- combined_data %>% group_by(Id) %>% summarise(AvgSteps = mean(TotalSteps, na.rm = TRUE), AvgSleepMinutes = mean(TotalMinutesAsleep, na.rm = TRUE), AvgSedentaryMinutes = mean(SedentaryMinutes, na.rm = TRUE), AvgSleepEfficiency = mean(SleepEfficiency, na.rm = TRUE))

12. Verification

• Verified cleaned data using summary statistics:

• summary(combined_data) provided min, max, mean, and median values for key metrics.

• Visualized relationships to confirm data consistency:

• Scatterplots: TotalSteps vs. SedentaryMinutes, TotalMinutesAsleep vs. TotalTimeInBed.

13. Saving Cleaned Data

• Exported cleaned datasets for further analysis:

• Combined data: write.csv(combined_data, "cleaned_combined_data.csv", row.names = FALSE)
• User summary: write.csv(user_summary, "user_summary.csv", row.names = FALSE)

Summary

The cleaning and manipulation process ensured the dataset’s integrity and consistency, enabling accurate analysis of consumer trends. Key steps included handling duplicates, aligning data formats, creating derived metrics, and verifying the results through visual and statistical checks. The cleaned and transformed data is now ready for insightful analysis to support Bellabeat’s marketing strategy.

Analysis Summary

Guiding Questions and Responses

1. How should you organize your data to perform analysis on it?

• The data was organized into two key datasets:

• Combined Dataset: Includes merged daily activity and sleep data, enriched with derived metrics like SleepEfficiency and ActivityToSedentaryRatio.
• User Summary Dataset: Aggregated metrics per user, such as average steps, average sleep minutes, and average sedentary minutes.

• Both datasets were structured to allow easy filtering and grouping for insights at individual and aggregate levels.

2. Has your data been properly formatted?

• Yes, the data was properly formatted:

• Dates were consistently reformatted as YYYY-MM-DD.
• Numerical columns were validated to ensure all values were non-negative.
• Derived metrics were calculated for enhanced analysis.
• Columns irrelevant to the analysis, such as redundant date fields, were removed.

3. What surprises did you discover in the data?

• A significant proportion of participants had low activity levels, with some reporting fewer than 1,000 steps on certain days, indicating potential sedentary behavior.
• Sleep patterns showed inefficiencies: while participants spent considerable time in bed, a notable portion had SleepEfficiency below 90%.
• There was an inverse relationship between high sedentary minutes and total steps, suggesting limited activity on days with prolonged inactivity.

4. What trends or relationships did you find in the data?

• Activity Patterns:

• Users with higher total steps generally had fewer sedentary minutes.
• Lightly active minutes contributed significantly to total activity, while very active minutes were minimal for most participants.

• Sleep Patterns:

• Sleep duration correlated strongly with TotalTimeInBed, but inefficiencies were observed in some cases where participants spent longer in bed but achieved less sleep.
• Median SleepEfficiency was around 94%, but some outliers indicated inefficiencies below 70%.

• Combined Insights:

• Participants with balanced activity and sleep routines displayed higher overall health indicators (e.g., fewer sedentary minutes and higher SleepEfficiency).

5. How will these insights help answer your business questions?

• Understanding Usage Trends: Insights into user activity and sleep behaviors help identify key customer segments (e.g., active users versus sedentary users).
• Product Enhancement: Sleep inefficiencies and low activity levels suggest opportunities to design features that encourage movement and improve sleep quality.
• Marketing Strategy: Trends highlight potential messaging strategies:

• Promote Bellabeat products as tools for improving daily activity and achieving better sleep routines.
• Target messaging to different segments (e.g., sedentary individuals or those with inefficient sleep habits).

Key Tasks

1. Aggregate Data:

• Grouped and summarized metrics at the user level using group_by() and summarise() in R.
• Calculated averages for steps, sedentary minutes, and sleep efficiency.

2. Organize and Format Data:

• Ensured uniform column naming, date formatting, and data type consistency.

3. Perform Calculations:

• Derived new metrics:

• SleepEfficiency = TotalMinutesAsleep / TotalTimeInBed
• ActivityToSedentaryRatio = TotalSteps / SedentaryMinutes

• Conducted statistical summaries and visualized relationships using scatterplots.

4. Identify Trends and Relationships:

• Discovered patterns correlating activity and sedentary minutes, and sleep duration and efficiency.

Deliverable: Summary of Analysis

• The analysis revealed that many users exhibit sedentary behavior and experience suboptimal sleep efficiency, highlighting areas where Bellabeat’s products can provide value.
• These findings will inform marketing strategies tailored to user segments and product improvements to address user challenges.

Sharing Insights: Supporting Visualizations and Key Findings

Guiding Questions and Responses

1. Were you able to answer the business questions?

• Yes, the analysis provided actionable insights into user trends for activity, sleep, and sedentary behavior. These findings can inform Bellabeat’s product development and marketing strategies.

2. What story does your data tell?

• Activity: Users exhibit a wide range of activity levels. A significant portion has low activity levels, which Bellabeat products can address by encouraging movement through reminders or gamification features.
• Sleep: While most users have good sleep efficiency (~94%), a subset experiences inefficiencies, indicating a need for features to monitor and improve sleep patterns.
• Combined Insight: Active users tend to have higher engagement and better overall wellness indicators, while sedentary users and those with poor sleep habits present opportunities for targeted marketing and product enhancements.

3. How do your findings relate to your original question?

• The findings align with the objective to understand smart device usage patterns. They provide insights into trends and behaviors that can shape Bellabeat’s marketing strategy and product positioning.

4. Who is your audience? What is the best way to communicate with them?

• Audience: Bellabeat executive team, including Urška Sršen and Sando Mur.
• Communication Strategy: A combination of clear visualizations and concise explanations tailored to business decision-makers. The focus is on actionable insights rather than technical details.

5. Can data visualization help you share your findings?

• Yes, visualizations like scatterplots, bar charts, and trend lines effectively highlight relationships, trends, and areas of focus.

6. Is your presentation accessible to your audience?

• Visuals and text are designed to be clear, concise, and jargon-free. Key points are highlighted, and supporting data is simplified for easy comprehension.

Key Tasks

1. Determine the Best Way to Share Findings:

• Develop a slide deck with visualizations and summaries for Bellabeat’s executive team.
• Include clear explanations of insights and recommendations.

2. Create Effective Data Visualizations:

• Activity Trends: A bar chart showing average steps across users to highlight varying activity levels.
• Sleep Patterns: A scatterplot displaying TotalMinutesAsleep vs. TotalTimeInBed with a trend line to emphasize sleep efficiency.
• Combined Analysis: A clustered bar chart comparing average SedentaryMinutes and LightlyActiveMinutes across user segments.

3. Present Findings:

• Use visual aids to narrate the data’s story.
• Highlight key insights, such as low activity among users, sleep inefficiencies, and correlations between activity and wellness.

4. Ensure Accessibility:

• Use contrasting colors and large, legible fonts.
• Provide textual descriptions alongside charts for clarity.

Deliverable: Visualizations and Key Findings

Summary of Recommendations

• Target Marketing: Focus on segments with low activity and poor sleep efficiency. Emphasize how Bellabeat products can improve these areas.
• Product Features: Introduce tools for activity encouragement (e.g., reminders, gamification) and sleep improvement (e.g., guided relaxation, analytics).
• Positioning: Promote Bellabeat as a comprehensive wellness solution, leveraging the insights to address pain points for users.

High-Level Insights and Recommendations

Guiding Questions and Responses

1. What is your final conclusion based on your analysis?
   Bellabeat can significantly improve its market reach by leveraging insights into user activity and sleep patterns:

• Activity Insights: A substantial portion of users exhibit low activity levels, with high sedentary minutes. Bellabeat’s products can target these users by encouraging movement through personalized reminders or challenges.
• Sleep Insights: Although most users have good sleep efficiency, a subset experiences inefficiencies. This presents an opportunity to promote Bellabeat as a solution for improving sleep quality.
• Combined Insights: Users with balanced activity and sleep routines tend to have better overall wellness, making them a key demographic for retention-focused strategies.

2. How could your team and business apply your insights?

• Marketing Strategy: Develop targeted campaigns highlighting the benefits of Bellabeat products in improving activity levels and sleep quality.
• Product Development: Enhance features in the Bellabeat app to include:

• Activity challenges and gamification for low-active users.
• Advanced sleep monitoring and personalized tips for improving efficiency.

• Customer Segmentation: Use data-driven segmentation to create personalized experiences for high-value user groups.

3. What next steps would you or your stakeholders take based on your findings?

• Immediate Actions:

• Design marketing content aimed at sedentary users and those with poor sleep habits.
• Develop app updates that focus on actionable insights for activity and sleep.

• Long-Term Actions:

• Collect and analyze more data to refine insights into user behavior.
• Track the effectiveness of new app features and marketing campaigns.

4. Is there additional data you could use to expand on your findings?

• Data on nutrition and stress levels: This could complement activity and sleep data for a holistic wellness strategy.
• Demographic information: To better understand user preferences based on age, location, and lifestyle.
• Competitive data: Insights into how users engage with non-Bellabeat devices to identify competitive advantages.

Key High-Level Insights

1. Target Sedentary Users:

• Insight: Many users log high sedentary minutes with low daily steps.
  Recommendation: Position Bellabeat products as tools to motivate movement and reduce inactivity.

2. Enhance Sleep Monitoring Features:

• Insight: Some users spend significant time in bed but achieve suboptimal sleep efficiency.
  Recommendation: Introduce advanced analytics and guided relaxation techniques in the app to improve sleep.

3. Segmented Marketing Campaigns:

• Insight: Different user segments (active vs. sedentary, good sleepers vs. poor sleepers) have varying needs.
  Recommendation: Use segmentation to tailor marketing efforts and personalized product suggestions.

4. Retention Through Wellness Tracking:

• Insight: Active and balanced users are more likely to stay engaged with the app and products.
  Recommendation: Develop loyalty programs or premium features that reward consistent usage and engagement.

Next Steps for Implementation

• Portfolio Creation: Compile this case study into your portfolio, highlighting the data cleaning, analysis, and actionable recommendations.
• Practice Presentation: Prepare and practice presenting this case study to stakeholders, focusing on insights and their business implications.
• Continuous Improvement: Gather feedback on your findings and recommendations, iterating based on additional data and stakeholder needs.