Playing smart by studying Smart Devices

Exploratory Data Analysis with R programming to provide insights about competitors's consumer behaviour

This is an optional capstone project from the Google Data Analytics Course.

No real stakeholders/clients were involved in it.

The business task

Bellabeat, a wellness company best known for their wearable jewelry, has rapidly positioned itself as one of the most popular health trackers for women. Their CCO thinks that an analysis of non-Bellabeat consumer data (FitBit fitness tracker) could reveal interesting trends - insights extracted from their competitors' usage data could create more opportunities for growth. The following business objectives are derived from this task:

What are the trends in user behaviour of FitBit tracker over the course of a month?

In what ways might these patterns be relevant to Bellabeat's customer base?

How could these trends guide the development of new features for Bellabeat's upcoming app version?

Key insights

Download R Markdown to see full analysis and associated code

No differences in activity throughout the month, but there is an obvious time window each day when users are more active.

Active minutes per day (Y axis) over one month (X axis).png

Users are very sedentary (+15 h/day inactive), but they reach the number of steps/day considered minimal for health benefits.

There is a correlation between high sedentarism time and short sleep time.

DATASET LIMITATIONS

Some analyses were jeopardized or impossible to perform because they depended on data requiring a manual input (weight, height, age, sex) that were not added by most users

Data gathered in 2016 might not align with current habits of fitness tracker users

Sample size of 24-33 FitBit users for one month of use isn't representative of the entire fitness population behaviours throughout the year

Since the data was collected through a survey, its reliability and accuracy was impossible to confirm

For the target population size (number of connected wearable devices worldwide according to statista), 1105M, and with a margin of error of 17%, the conclusions extracted from the data have a confidence level of 95%.

Data-driven recommendations

The baseline of daily steps and activity depends on multiple factors, according to a study from 2011. For every extra 1000 steps above baseline, the risk of all cause mortality and cardiovascular disease is reduced regardless of age, sex, health conditions or behaviours according to a study from 2022.

Adjust automatically the baseline of daily steps to the user’s age, sex, and health status (disability/chronic illness) (source).

For every extra 1000 steps above baseline, a push notification informs the user they have “increased their health benefits” or will "sleep better tonight".

Create a push notification by default with a motivational quote at 16:30 (right before peak activity hours) every day to get the user to work out.

To encourage users to input all necessary data for proper functioning of the built-in app calculations, it could be useful to create a step during initial app setup that emphasizes its importance.

For inclusivity and accuracy of estimations, include a manually input section in the profile about disability/chronic illness.

Recommend the user to get a medical check up through an in-app notification whenever outliers in average values of heart rate are detected.