Unlocking Better Sleep for Women: How to Leverage the Bellabeat Leaf for a Healthier Night's Rest

In the rapidly evolving world of health-focused products, Bellabeat has emerged as a successful player, catering to the wellness needs of women. As the company aims to solidify its position in the global smart device market, the power of data analysis becomes crucial.

As a junior data analyst on Bellabeat's marketing team, I was tasked with delving into smart device data collected by FitBit to gain valuable insights into consumer behavior. I will be focusing on Bellabeat’s classic wellness tracker, "The Leaf", which can be worn as a bracelet, necklace, or clip. The Leaf tracker connects to the Bellabeat app to track activity, sleep, and stress. In this article, I will explore the key findings from the analysis, highlighting the opportunities it presents for Bellabeat's marketing strategy.

Methodology:

In order to answer the key business questions, I will follow the six key steps of the data analysis process: ask, prepare, process, analyze, share, and act.

Ask:

Identify the business task: Harnessing publicly available FitBit Fitness Tracker Data from Kaggle, Bellabeat aims to uncover insightful trends in smart device usage. Through a comprehensive analysis, this data will enable Bellabeat to identify patterns that can drive its marketing strategies to new heights.

Prepare:

This FitBit Fitness Tracker Data | Kaggle data set contains a personal fitness tracker from 30 Fitbit users. The users consented to the submission of personal tracker data, including

1. Minute-level output for physical activity,
2. heart rate,
3. Weight log and
4. sleep monitoring.

To better understand the performance indicator of Bellabeat Leaf Tracker, we are choosing to explore and analyze the sleep data from the sleepDay_merged.csv file

The data source under consideration is evaluated based on the ROCCC criteria: Reliable, Original, Comprehensive, Current, and Cited.

Reliable: The data is rated as LOW in reliability since it consists of responses from only 30 female participants, which may not be representative of the broader female population.

Original: The originality of the data is rated as LOW as it was collected through a third-party provider and may not directly align with Bellabeat's specific research objectives.

Comprehensive: The dataset's comprehensiveness is considered HIGH as its parameters somewhat match those of Bellabeat's products, but there may still be limitations in capturing a complete picture.

Current: The data's currentness is rated LOW since it is collected over a month in 2016, making it potentially outdated and less relevant for present-day analysis.

Cited: The dataset is ranked as LOW in terms of citation since it was obtained from a third-party source, and the credibility or origin of the data is unclear.

Overall, this dataset is deemed of poor quality, and it is not recommended to base critical business recommendations on this data due to its limited reliability, lack of originality, incomplete scope, outdatedness, and unknown sourcing. A more robust and up-to-date dataset should be sought to derive meaningful insights for Bellabeat's marketing strategy.

Process:

We are using python packages numpy, matplotlib, datetime, pandas and seaborn to accomplish the data analysis and visualization.

• By looking into the data, I discovered that 24 out of 30 users have logged in their sleep records for a span of one month starting from April 12, 2016, to May 12, 2016.
• Data looks fairly clean with no missing values. However, the "SleepDay" has been wrongfully entered as "object". I will convert the 'SleepDay' column of the DataFrame data to a proper datetime object.

Analyze:

Now that we have data processed and nicely formatted for analysis, we will gain deeper insight through data visualization.

Let's visualize the number of sleep records for each user across different weekdays. It organizes the data to present it in a stacked bar chart where each user is assigned a unique and easy-to-read identifier. We present data for 24 users, and the resulting pivot plot provides a clear view of the sleep patterns for each user across different weekdays, making it easier to identify trends or anomalies in the data.

The illustration reveals that only half of the users have tracked their sleep records for more than the suggested 20 days. This limited data collection could potentially impact the accuracy of our results when identifying suitable sleep enhancement product recommendations for Bellabeat's clientele.

Next, we generate a stacked bar chart that shows the sleep quality for each user in the data. It does so by first categorizing the sleep records into different quality levels, then visualizing the number of records for each sleep quality level by user.

I have defined a custom function called sleep_quality_category to categorize sleep quality based on the total minutes of sleep. There are three categories: "Good sleep" (>= 480 minutes), "Average sleep" (360 to 479 minutes), and "Poor sleep" (less than 360 minutes).

# Define sleep quality categories based on TotalMinutesAsleep
def sleep_quality_category(minutes_asleep):
? ? if minutes_asleep >= 480:
? ? ? ? return "Good sleep"
? ? elif 360 <= minutes_asleep < 480:
? ? ? ? return "Average sleep"
? ? else:
? ? ? ? return "Poor sleep"


# Categorize sleep quality for each record
data["SleepQuality"] = data["TotalMinutesAsleep"].apply(sleep_quality_category)


# Group data by "Id" and "SleepQuality", and sum the "TotalSleepRecords" for each group
sleep_quality_counts = data.groupby(["User", "SleepQuality"])["TotalSleepRecords"].sum().unstack(fill_value=0)

n        

The analysis above provides an interesting perspective on users' sleep quality; however, a significant limitation must be considered. Individuals who did not track their sleep records for at least the recommended 20 days are grouped within the "poor sleep" category. This could inadvertently skew the data and lead to the erroneous assumption that a large proportion of users are experiencing poor sleep.

In reality, what might be labeled as "poor sleep" could actually be a reflection of insufficient data collection rather than an accurate representation of an individual's sleep quality. It emphasizes the importance of a well-defined data collection process and the need for caution when interpreting results.

We also look into histogram with a kernel density estimate (KDE) of the "Sleep Duration" data. The histogram shows how many hours of sleep the users got. Each bar represents the frequency or number of data points in each bin. The width of the bins is determined by the range of the data and the specified number of bins.

The sleep duration data seems accurate and does not indicate anomalies.

However, a better understanding of sleep patterns can be found by illustrating the relationship between sleep duration and total time in bed through a regression analysis. The regression line provides valuable insight to study sleep habits, recognize patterns, and draw insights that can lead to improved sleep quality and well-being. It's more than a mere correlation; it's a gateway to understanding individual sleep behaviors and the potential factors influencing them.

The following code employs both the Pandas library (implicitly) to manage data and Seaborn for visualization.

The data presents few outliers when users stayed in bed for longer hours compared to their actual sleep hours.

The Hexbin plot, which is employed to visualize the correspondence between sleep hours and total hours in bed, provides key insights, especially considering that half of the users in the dataset did not maintain tracking for the suggested duration. Here's what the analysis reveals:

Enhanced Data Visualization: Unlike scatter plots that might overlap, Hexbin plots provide a two-dimensional histogram that effectively captures the density of the correlation.

Insights into Sleep Efficiency: The areas where sleep hours closely match total hours in bed might point to higher sleep efficiency. This pattern could suggest that those users are falling asleep quickly and staying asleep throughout the night.

Guidance for Sleep Improvement: By understanding this relationship, tailored recommendations can be provided to improve sleep habits. For individuals who spend more time in bed than asleep, strategies to improve sleep efficiency may be valuable.

Share:

Based on the observations mentioned, it appears that there are significant opportunities and challenges in the data and product strategy for Bellabeat.

1. Data Inconsistency Issues:

The Fitbit data that Bellabeat intended to use for insights is inconsistent and incomplete, making it challenging to derive accurate conclusions.

This highlights the critical need for data integrity and accuracy in any analysis. Incomplete data can lead to misleading results, which could adversely affect business decisions.

2. Improving the Leaf Tracker:

The Leaf tracker may lack aesthetic appeal or functionality that motivates users to wear it and track their sleep consistently.

User engagement with wearable devices often hinges on design, ease of use, and perceived value. If the device is not appealing or does not provide clear benefits, adherence may drop.

Investing in user-centered design, adding features that provide real-time feedback, or offering incentives for consistent tracking could enhance user engagement.

3. Need for Customized Data Collection:

Utilizing a broad survey across Bellabeat's user population would provide more representative and relevant data for analysis.

Designing and conducting a user survey, possibly integrated within the Bellabeat app, would allow for more controlled and purposeful data gathering.

4. Expanding Product Offerings:

There's potential for Bellabeat to recommend parallel product lines tailored to users' sleep quality, considering various factors like location, geography, weather, and lifestyle.

Personalization is a significant trend in the wellness industry. Leveraging personalized data can lead to products and recommendations that resonate more deeply with individual users.

5. Collaborative and Ethical Approach:

Collaboration with users in the data collection and product development process is not only ethical but can enhance user trust and product relevance. Implementing transparent data practices and actively involving users in product development through focus groups or beta testing could create a more user-aligned and responsible brand image.

Act:

The analysis emphasizes the importance of consistent, accurate, and comprehensive data collection for effective health and wellness insights. Bellabeat has the opportunity to improve its products, services, and user engagement strategies based on these insights. By integrating more extensive user data and addressing specific user needs, Bellabeat can significantly enhance its offerings and provide a more personalized and effective solution for sleep enhancement.

If you've found this exploration into data analytics as exciting as I do, I'd love to connect with you. My background in physics has led me to discover a passion for unraveling complex data and translating it into meaningful insights. Whether you share similar interests or have questions about the analysis, please feel free to connect with me Tazkera Haque | LinkedIn. Let's continue this conversation and explore the fascinating world of data together!"