Fitness Tracker Usage Analysis
Fitness Tracker Usage Analysis
for Bellabeat
for Bellabeat
by David Lentz
by David Lentz
Updated: 5/6/2024
Use these tabs to navigate the report:
INTRODUCTION
INTRODUCTION
Bellabeat is a tech-driven wellness company designed for women founded in 2014 and headquartered in San Francisco. By 2016, the company had quick growth and globally launched campaigns and opened locations in Zagreb, Croatia and Hong Kong, China.
The Bellabeat fitness products include: Bellabeat Application, Leaf Tracker, Time Watch, Spring Water Bottle, and Bellabeat membership.
The focus of this analysis is to suggest opportunities to improve the specific fitness product Leaf tracker. The Leaf is a pendant that measures your sleep cycles, activity, breathing, and monthly cycles in conjunction with the Bellabeat application. Leaf can be worn as a bracelet, necklace, or clip.
Leaf Tracker Timeline:
2015: Bellabeat launched a wearable tracker called Leaf Nature, characterized by its nature-inspired design and versatility of wear.[1] It is a wearable fitness tracker for women that tracks activity, sleep, and stress resistance.[2] The same year, the company added period and pregnancy tracking to its accompanying application.[3]
2016: The company launched Leaf Urban,[4] the updated version of the previous Leaf tracker that featured a more modern, water-resistant design, and updated the corresponding Bellabeat application with additional mindfulness guidance.[5]
2018: The Leaf collection was later expanded to include Leaf Chakra.[6]
2019: Bellabeat partners with Swarovski to launch Bellabeat Leaf Crystal, created with Swarovski crystals.[7][8]
2021: Bellabeat partners with Poosh by Kourtney Kardashian for the Poosh Your Wellness Festival, where the company held a solo panel with the topic “Living by Your Cycle With Bellabeat”[9] led by Bellabeat’s co-founder Urška Sršen and Shayna Taylor, a holistic nutritionist.
2023: Bellabeat hosts a Wellness Summer Power dinner at Nobu in Malibu, featuring Olivia Culpo as the host and attended by celebrities including James Charles, Sophia Culpo, Aurora Culpo, Montana Tucker, Sarah Stage, and others.[10] Bellabeat partners with The Body Shop on the Beyond Skin Deep wellness bundle[11] that combines Bellabeat’s Ivy wearable device with The Body Shop’s products.[12]
The analysis task was given to Bellabeat’s Marketing Team to perform data analysis and present top recommendations to Bellabeat’s Executive Team.
BUSINESS TASK SUMMARY
BUSINESS TASK SUMMARY
Business Task Statement
Compare usage data of Bellabeat’s competitors’ similar smart devices and present data insights and marketing recommendations to the Bellabeat executive team.
Bellabeat Mission Statement
Shaping Women’s Health & Well-Being: Focusing on creating innovative health and wellness products for women, our mission is to empower women to take control of their health by providing them with technology-driven solutions that blend design and function.[13]
Key stakeholders:
The Executive Team currently includes these six people:
- Sandro Mur: CEO, co-founder, and responsible for finance, strategy, and growth
- Urška Sršen: CPO and co-founder
- Lejla Sertovic: Co-Founder
- Josko Zunic: Head of Growth
- Natalija Cvjetkovic: Business Development Manager
- Irena Kekelj: Vice President Security
ASKING PHASE
ASKING PHASE
SPECIAL NOTE: This report contains very highly detailed subject matter that would not typically be included to present to these key stakeholders, but is included to show the thought processes throughout this case study, since it may be used to show future employers of my work capabilities.
The following questions were brainstormed by Bellabeat’s Marketing Team in order to steer the analyses. Keep in mind not all questions will be answered by this analysis, but may require other data analyses in order to answer those remaining questions. It was used as a general guide to lead the Marketing Team.
- What are some trends in smart device usage data patterns?
- How could these trends apply to Bellabeat customers?
- How could these trends help influence Bellabeat marketing strategy?
- Who are Bellabeat’s competitors?
- What devices do these competitors sell?
- What features are used?
- Why do women choose the Leaf tracker?
- Why do women choose other competing company’s fitness tracker devices?
- Are there features that could be added to Leaf tracker?
- Do men use smart device trackers?
- How is usage compared between men and women?
- Is Bellabeat open to selling devices for men?
- Is the data presenting other questions that should be asked?
- How are the current sales of Leaf trackers?
- Are Leaf trackers more popular or less popular than the watch Time, water bottle Spring, and the Bellabeat app?
- Do women buy more than one device?
- Do women use the app or just wear the Leaf?
- Can Leaf be used without the app?
- Do women that use the Leaf have memberships?
- Do women that have memberships use the features, i.e. personalized guidance on nutrition, activity, sleep, health, beauty, and mindfulness?
- Which features are most popular? Which are barely used?
- Are there other features that should be included on the app?
- Where did women learn about the app?
- Which is the most popular advertising method?
- What were the sales growth since 2014?
- What are future projections of sales growth?
- Do women find Leaf easy to use?
- Are there any glitches or problems reported by users?
- Which online retailers sell the products?
- Which online retailers sell the most products?
- Do people buy from the e-commerce channel on Bellabeat’s own website?
- How do these sales compare to other online retailers?
- Which advertising method does best: radio, out-of-home billboard, print, television, or digital marketing?
- Which advertising has Bellabeat focused on the most?
- Which digital marketing is most successful: Google Search, Facebook, Instagram, Twitter, YouTube video ads, or Google Display Network display ads?
- What are the support campaigns?
- Which support campaigns are most successful?
- What are the key marketing dates?
- What are opportunities for growth?
- How are people already using the Leaf tracker?
- What are high-level recommendations of how these trends can inform Bellabeat marketing strategy?
- Should the study include more than 30 users as a sample size?
- Where is the data from?
- Is the data validated?
- Is the data clean?
- Are there other miscellaneous questions generated during analysis?
PREPARE PHASE: DATA SOURCE DETAILS
PREPARE PHASE: DATA SOURCE DETAILS
There are 4 following datasets: the first was supplied by a key stakeholder and Chief Product Officer Urška Sršen. The second, third, and fourth sets were added to include more valid data, leading to more analysis, visualizations, insights, and recommendations beyond the first set.
DATASET 1:
FitBit Fitness Tracker Data: Pattern recognition with tracker data: Improve Your Overall Health
The FitBit Fitness Tracker Data is a secondary, external, discrete, quantitative, structured, open source data downloaded from Mobius at https://www.kaggle.com/datasets/arashnic/fitbit and organized in rows and columns in long format.
Fitness tracker data was measured and collected from 30 consenting Fitbit users spanning 2 months from March 12 to May 12 of 2016. The data downloads in a zipped file of 2 separate month folders, containing 29 .csv files. Data was collected by Amazon Mechanical Turk via survey responders and the data had no identifiable personal information included. The data is cited properly.
The personal data includes hour, minute, and second information on sleep, physical activity level, activity intensity, calories and metabolic equivalents per activity, weight, and measured steps.
The data is rated for CC0: public domain use.
This dataset contains substantial bias and limitations:
This data is not completely reliable because the data and its description do not mention gender.
The data is original, but not completely comprehensive, since the sleep file is only collected for the second month set.
The data is repetitive, including wide and long formats for the same data.
There are inconsistencies in the time elements of files. Some are reported in seconds, minutes, days, or a combination.
There are missing values in some of the data columns.
The data is not very current. It is from 8 years ago, in 2016.
The description of the data states the sample size is 30 participants. However, there are 35 distinct user IDs within the data sets.
The sample size is small at 35 and may contain men or women. It is not specified, whereas Bellabeat primarily focuses on women consumers. This is a bias that is noted.
The data is split up into separate months, each in their own folder, that will have to be combined or joined to analyze.
The data collection starts in the middle of the months, so this makes naming more challenging.
Seven .csv files are missing from the Mar 12 to Apr 11 folder that are contained in the Apr 12 to May 11 data file folders. Of the seven missing files:
Three “daily” files can be calculated using their counterpart “hourly” files.
Three “wide” files have already been converted to more easily used “narrow” files, so the wide formats can be ignored, deleted, or archived.
The 7th file is sleepDay_merged.csv and is only in the second month set.
- The term “narrow” is used in place of “long” in file names.
The primary data will require substantial cleanup documented in the PROCESS PHASE: DATA CLEANING AND TRANSFORMATION DOCUMENTATION section.
DATASET 2:
Replication Data for: Using machine learning methods to predict physical activity types with Apple Watch and Fitbit Data using indirect calorimetry as the criterion.
This data is secondary, external, discrete, quantitative, structured, open source data download from https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/ZS2Z2J organized in long format.
The study included 26 women and 20 men consenting and wearing either Fitbit or Apple Watch fitness trackers to obtain caloric expenditure data to see if the trackers were accurate. Participants were tested by 3 different intensities or self paced for 40 minutes on a treadmill, and then 25 minutes of either standing, lying, or sitting. The study concluded that Fitbit and Apple Watch were able to predict physical activity type with reasonable accuracy. As far as privacy, the data had no identifiable personal identification.
The data was rated for CC0: Public Domain usage Version 1.
This dataset has some limitations:
The sample size was small with only 46 participants.
The study doesn’t specify locations of people’s data or country of origin.
DATASET 3:
DATASET 4:
Replication data for Using Fitness Trackers and Smartwatches to Measure Physical Activity in Research
This data is a secondary, external, discrete, quantitative, structured, open source data download by UiT The Arctic University of Norway at DataverseNO https://dataverse.no/dataset.xhtml?persistentId=doi:10.18710/6ZWC9Z organized in long format.
The study consists of 423 types of fitness trackers or watches measuring which features each brand and type included. The country of origin, year released, company names, and device names are included. The data listed Boolean style arguments on features including an accelerometer, gyroscope, magnetometer, barometer, GPS, and PPG. Refer to glossary for technology definitions. Privacy was not an issue.
The data was rated for CC0: Public Domain usage Version 1.
The data was included in order to find trends globally and examine technology features in fitness trackers.
This dataset has some limitations:
- This dataset focused more on technology and features of fitness trackers and watches instead of fitness consumers. It may still give insight into usage and global statistics.
PROCESS PHASE: DATA CLEANING AND TRANSFORMATION DOCUMENTATION
PROCESS PHASE: DATA CLEANING AND TRANSFORMATION DOCUMENTATION
DATASET 1: Fitbit Fitness Tracker Data
Extensive data cleaning was required in order to analyze this data set.
SPECIAL NOTE: This report contains very highly detailed subject matter that would not typically be included to present to these key stakeholders, but is included to show the thought processes throughout this case study, since it may be used to show future employers of my work capabilities.
The original data was copied to a RAW folder and locked with a password so no changes could be made. It was then recopied to a CLEANED folder to begin the scrubbing process. The RAW folder files will stay in their original form in case a backup replacement is needed. The file names will be consistent and standardized utilizing Bellabeat’s naming convention policy.
“Narrow” will be changed to “long”. “Daily” will be changed to “day” for time element consistency.
All file naming conventions will be changed to snake case, using lowercase and underscores to separate the individual file words.
“Merged” and “narrow” can all be discarded since they are no longer needed and all data is in long format.
Standardizing all time elements to plural and removing the “ly” from them shows consistency. All time elements in file names will be moved to after the main element (i.e. “hourlyCalories_merged.csv” will become “calories_hours.csv”.
METs will change to metabolic_equivalents, which are a way to measure the energy cost of physical activities, and are used by Fitbit devices to estimate exercise intensity.
The file WeightLogInfo will be changed to include the unit of measure kg to remain consistent (weight_kg.csv). The LogInfo column will be ignored because it does not contain useful information.
A folder will be created to contain the merged data from both months, named fitbit_data_03122016_05122016.
Google Sheets will be used to merge files and then column names only have to be standardized once instead of twice to save time. A “_month1” and a “_month2” will be added to the end of file names in the 1st and 2nd month’s folder, respectively, so the files with the same names wouldn’t be confusing during the merge. The cut and paste functions were used in Google Sheets to merge the files. Heartrates by seconds were 2 files that were too large to merge, so Google Cloud Services were activated in order to upload and merge those files.
Calories were in both hourly and in minutes. The minute increment was too small and hourly or daily calories would be more useful, so the calories_minutes files will be ignored, deleted, or archived depending on policy.
Google Sheets was used to merge all files, since Sheets can accommodate up to 5 million rows.
Google sheets was used to trim spaces, delete replicated lines, and utilize conditional formatting to change empty or null cells to zeros when warranted. Sorting and filtering assisted the cleaning process. Outliers were examined and discarded on an individual basis as per dataset logic.
The date_time columns were formatted as Date and Time datatypes to ensure consistency and for future JOINing functions in Tableau.
Variable names of columns and/or rows were standardized using snake case as per Bellabeat’s policy on naming conventions.
The cleaned files were then saved to their respective “_CLEANED” folders in .csv format to easily be uploaded to Tableau, BigQuery, or RStudio for analysis.
All files were downloadable as .csv files and needed no other conversion for consistency.
These prior steps helped transform the original data into high quality cleaned data that ensured data validation standards.
DATASET 2: Apple Watch and Fitbit data
- The same steps mentioned above were repeated to ensure data validation. However, this data was pre-cleaned and only had three data tables.
DATASET 3: Exercise and Fitness Metrics
- The same steps mentioned above were repeated to ensure data validation. However, this data was pre-cleaned and only had one data table.
DATASET 4: Fitness Trackers and Smartwatches
The same steps mentioned above were repeated to ensure data validation. However, this data was nearly cleaned and only had one data table.
One data value had to be cleaned because it was a company name made of only numbers, so it was a misread datatype of integer instead of string text. It was converted to string text for consistency.
ANALYSIS PHASE 1: ANALYSIS SUMMARY
ANALYSIS PHASE 1: ANALYSIS SUMMARY
Google Sheets - The dataset files were imported into Google Sheets. The Id columns were all the same, but the Date or Time columns had many different formats. The datatype was changed to Date and Time with column label “date_time” in order to create a consistent format (i.e. 4/12/2016 08:35:00) and join tables later. Merging the 2 months files together was attempted in Sheets, but the larger files kept crashing. All files were exported into .csv files in a CLEANED folder. Many pivot tables were created in Google Sheets to consolidate data for specific users, averages, sums, minimums, maximums, counts, and distinct counts.
Text Editor - Merging the 2 separate month files into one file was done in a simple text editor utilizing cut and paste. The files were renamed with “_both_months” at the end of file names. This created less data tables in Tableau to join later. The files were saved as .csv files. “Weight_kg_month1” and “Sleep_day_month2” could not be merged with their counterparts because they did not exist in the datasets. They will still be imported to see if any insights can be determined from them.
Tableau - The 9 consolidated datasets were imported into Tableau. The “Activities_day_both_months” table was used as the main table to connect all other relational tables. Id and date_time were the common links used as primary keys to join all the tables. The joined tables in Tableau appear as follows:
Google Sheets, Tableau, and SQL in BigQuery were used to organize and analyze the data in many different ways and multiple outputs and visualizations were created to demonstrate the findings.
Some of the following SQL queries were used in BigQuery to organize tables into manageable summary tables:
- Example 1:
SELECT Id,
COUNT(Id) AS ID_count_entries,
SUM(TotalSteps) AS SUM_TotalSteps,
SUM(TotalDistance) AS SUM_TotalDistance,
SUM(TrackerDistance) AS SUM_TrackerDistance,
SUM(LoggedActivitiesDistance) AS SUM_LoggedActivitiesDistance,
SUM(VeryActiveDistance) AS SUM_VeryActiveDistance,
SUM(ModeratelyActiveDistance) AS SUM_ModeratelyActiveDistance,
SUM(LightActiveDistance) AS SUM_LightActiveDistance,
SUM(SedentaryActiveDistance) AS SUM_SecondaryActiveDistance,
SUM(VeryActiveminutes) AS SUM_VeryActiveMinutes,
SUM(FairlyActiveminutes) AS SUM_FairlyActiveMinutes,
SUM(LightlyActiveminutes) AS SUM_LightlyActiveMinutes,
SUM(SedentaryMinutes) AS SUM_SedentaryMinutes,
SUM(Calories) AS SUM_Calories
FROM `bellebeat-421514.health_data.activity`
GROUP BY Id- Example 2:
SELECT COUNT(DISTINCT(Id))
FROM `bellebeat-421514.health_data.weight`- Example 3:
SELECT * FROM `bellebeat-421514.bellebeat_data.heartbeats_seconds` CROSS JOIN `bellebeat-421514.bellebeat_data.heartrate_seconds2`RStudio - An R Markdown report (this report) was generated in RStudio to document the entire analysis process from beginning to end and saved as HTML output for an online portfolio that will be created to share the information with key stakeholders and fellow Marketing Data Analysts.
Kaggle and Linkin will provide the web space for uploading the final report to a portfolio.
ANALYSIS PHASE 2: VISUALIZATIONS AND KEY INSIGHTS
Most of the visualizations are from the primary dataset due to the multiple datasets. The three other datasets added to the case study had one table each to analyze, so there are less visuals and insights.
DATASET 1: Fitbit Fitness Tracker Data Visualizations
The following visualizations showed insights as noted below the images:
This is one of the more relevant visualizations showing what percentage of the 35 users actively used each of the 9 features listed. Five of the 9 features were utilized by all 35 users, including activities, calories, intensities, METs, and steps. Four of the 9 features were only partially used, including sleep times, sleep daily monitoring, heartrates, and weights from most used to least, respectively.
This image shows the average values for body fat and BMI per each user ID. Only 2 people used the Fitbit to calculate body fat, and only 11 used it to calculate BMI. The total users in the study were 35.
This 3 images above indicate 25 of 35 users utilized the Fitbit to track sleep time patterns, analyzing minimum, average, and maximum sleep times. The maximum amount of time spent sleeping in bed for people is: minimum 58 min (approx 1 hr), average 419 min (approx 7 hrs), and maximum of 796 min (approx 13.3 hrs). This is quite varied for the 25 users. The average amount of time in bed that it takes to actually fall asleep is quite a similar gap for most people, with the average being 39.3 minutes to fall asleep.
This visual showed that of the 25 out of 35 users, most of them used it often, at least one month or longer. The one quarter that tried the feature barely used the sleep tracking for long (less than 3 days).
When examining linear regression, the R² values are quite high in the first 3 graphs and very low on the fourth graph. These first 3 images show a strong correlation between very active minutes and very active distance, moderately active minutes and moderately active distance, and finally lightly active minutes and lightly active distance. These correlations make sense. However, sedentary minutes versus sedentary distance strongly does NOT correlate. Perhaps this is due to a sedentary person on the couch not moving much at all versus being a sedentary person in a traveling car over a certain distance.
All 35 users had data for the 4 types of distance each performed. This graph shows the sums of each type of distance: very active, moderately active, lightly active, and sedentary distances per each user. Three users traveled very active distances, but the majority traveled lightly active distances.
All 35 users had data for the 4 types of minutes each performed. This graph shows the sums of each type of time: very active, moderately active, lightly active, and sedentary minutes per each user. The sedentary minute sums are way higher than active minute sums, but of the 3 active time types, the majority were lightly active.The graph shows that it is harder to maintain very active and moderately active actions for extended periods of time.
This graph shows that 3 of 35 users actively logged their activities distance manually instead of automatically. This is a largely UNUSED feature.
This graph shows minimal correlation between the calorie sums and the total step sums. This was a surprising finding, because it was hypothisized that there would be a strong relation between the two variables. So more total steps taken do not necessarily mean one will burn more calories.
This image shows data for only 11 of 35 users. It shows 9 users manually tracked data and only 2 used automatically collected data. It is unsure as to why the other 24 did not have automatically collected data, or if it wasn’t presented in the data, or that feature can be turned off. These features are hardly used, because 9 of the 11 only had one point of data, meaning they each only tried it once.
This shows that 11 people out of 35 used the weight measurement feature. There is very little weight differences between minimum, maximum, and average per user, meaning nobody gained or lost much weight in 2 months time. Weight is in kilograms.
This grouped all 11 weight feature participants together with minimum, average, and maximum weights in pounds, but separated into 2 months. The minimum weights dropped slightly over time. The average weights increased slightly over time. The maximum weights increased significantly over time. The minimum and average BMIs dropped slightly over time, but the maximum BMIs increased significantly over time. It was easier to see trends using pounds versus kilograms, probably due to larger numbers in pounds vs. kg.
This graphs shows 15 users recorded heartrate measurements. Some of the maximum heartrates approach dangerous territory when above 200 bpm while working out. Some minimum rates can approach danger at the low end near 30 bpm. It may mean the electrical conduction of your heart is not firing properly. Age would have to be factored in to know the dangerous heartrate levels per each individual.
The 2 graphs above show linear regressions to discover correlations between types of distances and calories. Lightly active distance and very active distance shows slightly stronger correlations to calories than the moderately and sedentary. The second graph was created to show the 2 higher correlations separated out. However, none correlate very well at all.
The graph above shows linear regressions to discover correlations between types of minutes and calories. Lightly active distance and very active time shows stronger correlations to calories than the moderately and sedentary. However, none correlate too well at all.
DATASET 2: Apple Watch and Fitbit Data Visualizations
The first graph show the study’s general breakdown of female vs. male users. The second shows the distribution of activity types studied. The sample size is 46 users total. The distribution of both is relatively equal measures, which is best practice to aquire data in a study.
This graph shows a breakdown of counts of activities per gender. Men performed the study activities slightly more than women even though there was nearly equalivalent women and men studied.
This graph shows a break down of average dream weights (kgs) versus average actual weights per gender. The dream and actual weights were never too far off each other. This was a surprising finding due to the hypothesis that people typically have unrealistic ideal weights in mind for themselves. People were more realistic than was presumed. The ages spanned from 18 to 60 years old for participants, but were not correlated at all to ideal or actual weights.
This graph shows the average calories for each of the seven activities in the study. It was surprising that running 7 METs was closer to sitting and lying than compared to the other 3 activities of self-paced walk, or running 3 or 5 METs. The latter 3 activities burned the most calories.
DATASET 3: Exercise and Fitness Metrics Data Visualizations
Weather doesn’t seem to affect if people exercise or not. The data description didn’t elaborate on what the ten exercises were or if they were performed indoors or outside. If the data was collected inside, which would make sense, then the weather would not correlate. So this graph probably indicates the exercise measurements took place indoors, because it is hypothesized that if exercises were performed outside, then weather, especially rainy, would have shown an effect. Even though this data study did not show any proven insights from the analysis, the feature of including weather data in a fitness tracker could be a useful idea for outdoor fitness activists.
The exercise intensity averages seemed to be similar for all ten exercises. Unfortunately, the study did not describe what each of the ten exercises were.
DATASET 4: Fitness Trackers and Smartwatches Visualizations
This graph show the study’s general breakdown of Fitbit trackers vs. Apple Watch trackers. The sample size is 423 users total, which is quite large.
This image shows the technology inside 423 fitness trackers. Of the 423 devices, they all contain accelerometers. The other technologies are not used often. The lesser used technology features may be something to investigate farther to see if they could add benefits to current fitness tracker technologies. See GLOSSARY for technology definitions and uses.
This image indicates the percentage of the top producers of fitness trackers in the world. The US and China are the top producers.
ABBREVIATIONS
ABBREVIATIONS
- BMI - body mass index
- CC0 - Creative Common Zero
- CEO - Chief Executive Officer
- CPO - Chief Product Officer
- .csv - comma separated values
- GPS - Global Positioning System
- METs - Metabolic Equivalents
- PPG - Photoplethysmography
- R² - R-Squared (coefficient of determination) is a statistical measure in a regression model that determines the proportion of variance in the dependent variable that can be explained by the independent variable. In other words, r-squared shows how well the data fit the regression model (the goodness of fit).
GLOSSARY
GLOSSARY
accelerometer - An accelerometer is a sensor that measures the rate of change of velocity, or acceleration, in one or more directions. In fitness trackers, accelerometers convert body movement into data, such as steps taken, distance traveled, and calories burned. Accelerometers can also be used to identify postures, classify movements, estimate energy expenditure, and detect falls.
barometer - A barometer is a sensor that measures air pressure. In a smartwatch, a barometer is used to measure the altitude, provide weather forecast, and track the changes in elevation while climbing or hiking.
BMI - Body Mass Index is a number that estimates body fat based on a person’s height and weight. BMI is calculated by dividing a person’s weight in kilograms by the square of their height in meters. A high BMI can indicate high body fatness.
ECG Sensor - Electrocardiogram (ECG) sensors, also known as electrodes, are devices used to visualize the electrical activity of the heart. An ECG, or EKG, measures the voltage changes that occur on the skin as electrical signals travel from the heart’s pacemaker to the upper and lower chambers. The ECG tracing shows how the depolarization wave moves during each heartbeat, which is a wave of positive charge. The pattern of the wave depends on the set of electrodes used. For example, one electrode might be placed on the right arm and the other on the left leg, and a positive deflection will occur when the waves move toward the left leg electrode.
GPS - Global Positioning System, is a global navigation satellite system that provides location, velocity and time synchronization.
gyroscope - A gyroscope is a sensor that measures angular velocity. In a smart device, the gyroscope is used to track the orientation of the device, enabling features such as screen rotation and step counting.
heart rate monitor - Heart rate monitors are devices that can detect and track your heart or pulse rate continuously. Most of these devices are wearable, and many are highly accurate.
magnetometer - A magnetometer in a smart device is a type of sensor that measures the direction and strength of a magnetic field.
METs - One MET is the energy you spend sitting at rest — your resting or basal metabolic rate. So, an activity with a MET value of 4 means you’re exerting four times the energy than you would if you were sitting still. To put it in perspective, a brisk walk at 3 or 4 miles per hour has a value of 4 METs.
PPG - Photoplethysmography, which is a non-invasive optical measurement technique that uses a light source and a photodetector to measure blood volume variations. It is often used for heart rate monitoring in fitness, professional sports, and off-hospital monitoring. PPG sensors are cheaper and use simpler hardware than traditional ECG-based systems.
skin temperature sensors - Skin temperature sensors are devices that measure the temperature of the skin’s surface and can be used to indicate body temperature. They can be placed on the skin at any site, but are often used on the forehead, axilla, or other areas. Skin temperature sensors can be used for biomedical temperature monitoring, efficacy testing, clinical research analysis, and more.
SpO2 sensor - or pulse oximeter sensor, is a small clip that measures the percentage of oxygen in a patient’s red blood cells. It’s a non-invasive device that uses a probe, red and infrared light sources, and photo detectors to transmit light through a patient’s blood. The device then measures how oxygenated and deoxygenated hemoglobin absorb light differently.
CITATIONS
CITATIONS
Perez, Sarah (September 30, 2014). “Bellabeat Debuts A Trio Of Health-Tracking Products For Moms-To-Be And Beyond, Including Jewelry, A Monitor And A Smart Scale”. TechCrunch. Retrieved December 1, 2022.
“Bellabeat Debuts LEAF, A Smart Charm Necklace That Monitors Your Stress Levels And Ovulation Schedule”. Bustle. May 28, 2015. Retrieved December 1, 2022.
Perez, Sarah (January 16, 2015). “Bellabeat Leaf, A Health Tracker For Women, Will Now Track Your Reproductive Cycle Too”. TechCrunch. Retrieved December 1, 2022.
Perez, Sarah (July 26, 2016). “Bellabeat Leaf’s new wearable and app help women tackle stress”. TechCrunch. Retrieved December 1, 2022.
Marie Claire (June 9, 2015). “6 Pieces Of Wearable Tech To Destress Your Mind And Make You Feel Calmer”. Marie Claire UK. Retrieved December 1, 2022.
Jovin, Ivan (September 21, 2018). “Bellabeat launches Leaf Chakra, a wearable for mind-body balance”. Gadgets & Wearables. Retrieved December 9, 2022.
KAPFUNDE, MUCHANETA (November 8, 2019). “Bellabeat’s New Leaf Crystal Designed To Help You Make Better Lifestyle Choices(FashNerd)”. techlovesstyle (in Chinese). Retrieved December 9, 2022.
“The new Bellabeat fashion accessory successfully manages stress - HAMAG BICRO” (in Croatian). Retrieved December 9, 2022.
Living By Your Cycle with Bellabeat: PYW 2022 | Poosh, retrieved December 9, 2022.
“Bellabeat Ushers in Summer with a Stunning Private Dinner, Hosted by Olivia Culpo and Isabel Alysa Vita at the Elite Nobu, Malibu”. www.flaunt.com. Retrieved September 26, 2023.
Weekly, La (July 31, 2023). “Bellabeat and The Body Shop Collaborate to Deliver a Holistic Wellness Experience - LA Weekly”. www.laweekly.com. Retrieved September 26, 2023.
Journal (July 11, 2023). “Strateška suradnja za novu eru wellnesa: Bellabeat se udružio s The Body Shopom i predstavio novitete”. Journal.hr (in Croatian). Retrieved September 26, 2023.
Bellabeat. (n.d.). About Us. Retrieved from https://bellabeat.com/about-us/
Images compiled from Google Chrome image search.
APPENDIX I: OTHER NOTES
APPENDIX I: OTHER NOTES
The following code was used to create the data tabs:
# {.tabset}The following packages were installed to use R Markdown:
install.packages("rmarkdown")
install.packages("RStudio/Pandoc")
install.packages("LaTeX")
library(rmarkdown)
library(RStudio/Pandoc)
library(LaTeX)SPECIAL NOTE: This report contains very highly detailed subject matter that would not typically be included to present to these key stakeholders, but is included to show the thought processes throughout this case study, since it may be used to show future employers of my work capabilities.
Thank You for Examining my Case Study! All work contained in this report by David Lentz. THE END.
APPENDIX II: OTHER VISUALIZATIONS WITH LESS INSIGHTS, BUT CREATED FOR ANALYSIS EXPLORATIONS
APPENDIX II: OTHER VISUALIZATIONS WITH LESS INSIGHTS, BUT CREATED FOR ANALYSIS EXPLORATIONS
1. Fitbit Fitness Tracker Data
This image shows no discernible pattern to average types of calories over time.
The 4 previous graphs have lower R² values in their linear regression, showing minimal correlations between the compared variables. They correlate less in descending order. It was a surprise that maximum steps and maximum calories had the lowest correlation because it was hypothesized that those variables would have a strong connection.
This visual showed individual users and their plotted sleep states over time. A better chart was created to show usage in the ANALYSIS PHASE 2: VISUALIZATIONS AND KEY INSIGHTS section above.
This image was a little too busy and separated out to see visually better, as presented in the ANALYSIS PHASE 2: VISUALIZATIONS AND KEY INSIGHTS section above.
This image is too busy, but includes the minimum, average, and maximum times in bed, asleep, and not asleep. It was split into 3 other images to make it more presentable and shown above in the ANALYSIS PHASE 2: VISUALIZATIONS AND KEY INSIGHTS section.
Values were missing in this graph, so it was recreated in Tableau to show all values shown in the ANALYSIS PHASE 2: VISUALIZATIONS AND KEY INSIGHTS section.
2. Apple Watch and Fitbit data
These 2 graphics were intended with the best intentions to compare Fitbit to Apple Watch side by side. But after analyzing, it was realized they were poor quality graphed results and discarded due to faulty computations. The heart rate averages were too low to be realistic.
3. Exercise and Fitness Metrics
There were no unused visualizations in this study.
4. Fitness Trackers and Smartwatches
