CaseStudy BellaBeat

Bellabeat Case Study

What are the trends?
How does it apply to Bellabeat customers?
How will it influence Marketing Strategy?

Data I pulled is stored on Kaggle’s public domain by Mobius.
I’ve downloaded both sets to my hard drive.
Data is organized in long format typically 3 columns wide with a consistent
ID, Activity Date and then the Particular Data pulled for each CSV file.
30 days of data is pulled twice 1 month a part.
1st set of Fitabase Data 3.12.16-4.11.16 has 11 csv files and 2nd set of Fitabase data 4.12.16-5.11.16 has 19 csv files totaling 29 files.

Key Differences between sets:
1st Set is missing:
dailyCalories_merged
dailyIntensities_merged
dailySteps_merged
minuteCaloriesWide_merged
minuteIntensitiesWide_merged
minuteStepsWide_merged
sleepDay_merged
Cleaned in Excel
Charted in Excel and RStudio
At first glance I determined the data still needs to be cleaned due to inconsistencies.
My first thought was to see how many unique ID’s there were by filtering the ID column.
The 1st file I opened was the dailyActivities_merged in the 3.12.16 set. It had 37 unique ID’s. The 4.12.16 set had 35 unique ID’s.
As previously mentioned, and after opening many of the files there is an inconsistency with the amount of data provided. Another example is the Weight Log Info. Only 11 from the 1st set and 8 from the second set provided their weight log. Half of the Weight Log data was manually input vs the Data pulled from the actual device which can easily assert bias.
My 1st assumption is that for a company this large, only having 35 users sampled and 2 months of 3rd party data does not give BellaBeat adequate data and can easily assert bias.
Where and who were these 34 samples taken from?
Where they randomly sampled from different climates or terrain?
Where can we receive a larger data sample?

I’ve formatted the data from the dailyActivities_ merged sets into 5 columns and deleted any data that isn’t complete. The 5 columns will now be ID, Steps, Distance, Minutes and Calories. Do to inconsistencies between ID’s I could only use 29 ID’s.

Opinion: One thing that stands out is that people seemed to be more active the second month. This may be due to the more you exercise the easier it gets. There is a clear correlation between distance/steps with calories burned. There is also a clear correlation between minutes and calories burned when minutes of all intensities are combined. This can provide hope for those who want to be more intense but physically can’t. If they put in the time even with low intensity, they can still reach their goals.
Also, I’d like to find out what are the major complaints on reviews of different devices. What’s missing? Comfort? Ease of use, etc...
So this is definitely biased data but I wouldn’t hesitate to provide my findings on many of the complaints that competitors receive that might help BellaBeat beat the competition.
A list of most common 1-star complaints:
Data inconsistencies because of sweat. Selling a band that can absorb sweat is a must, and I recommend making the bands washable.
Data privacy assurance
Bugs from updates
Appearance
Accuracy such as steps and calories during resting state.
Sleep log Accuracy.
Sync issues especially with Bluetooth.
GPS accuracy
Battery life
Customer Support
Comfort.

In Conclusion, BellaBeat needs to focus on using their own data rather than a small sample from a competitor. I don’t see any data from this dataset that can give value and answer the questions that are really needed to help BellaBeat with marketing. The only take away from the data is that 25 out of the 29 users increase their steps and distance the second month. This can be a slogan of some sort for BellaBeat. I believe BellaBeat is better off listening to their own customers’ needs and complaints and exploring the reasons why their competitors are losing customers.