Bellabeat Case Study
Marie-Joe Kallassy
2022-06-02
Introduction
The following case study is the final capstone project in the Google
Data Analytics Professional Certificate. The case study is on Bellabeat,
a high-tech wellness company that manufactures health-focused products
made specifically for women. We are asked to analyze smart
devices-fitness data to gain insight into how consumers are using their
smart devices, and to discover trends that can guide a marketing
strategy for the company to grow.
For this case study, I will be following the 6 steps of the data
analysis process: Ask, Prepare, Process, Analyze, Share, and Act.
1-Ask:
The business task:
Analyze Fitbit users’ data to find trends and insights into how consumers use their smart devices.
Key stakeholders:
-Urska Srsen: Co-founder and CCO of Bellabeat.
-Sando Mur: Mathematician and Co-founder of Bellabeat.
-Bellabeat marketing analytics team: A team of data analysts
that uses their analysis to guide the marketing strategy.
2-Prepare:
- The data I’m gonna use for my analysis is stored on the website
Kaggle (Kaggle: Fitbit
Fitness Tracker Data). It’s an open-source data, made available
through Mobius.
- The data, which is in long format, was collected from a survey via
Amazon Mechanical Turk, between 03/12/2016 and 05/12/2016, where 30
eligible Fitbit users consented to the submission of their personal
tracker data.
- Since the data only has 30 participants, it can’t be representative
of the population as a whole, and might cause a sampling bias.
The data has also been collected several years ago (in 2016), which means it’s not current, and might not be reliable to accurately drive business decisions in the present.
3-Process:
Even though the survey only has 30 participants, some of the data is
still too big to use in Microsoft Excel, since it’s represented in long
format.
So I decided to use R to clean and analyze the data.
I’ve also decided to use the following datasets in my analysis:
daily_activity, heartrate_seconds, and
sleep_day.
1- Loading all the packages needed for the analysis (already installed using install.packages():
library(readr)
library(tidyverse)## -- Attaching packages --------------------------------------- tidyverse 1.3.1 --## v ggplot2 3.3.6 v dplyr 1.0.9
## v tibble 3.1.7 v stringr 1.4.0
## v tidyr 1.2.0 v forcats 0.5.1
## v purrr 0.3.4## -- Conflicts ------------------------------------------ tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag() masks stats::lag()library(ggplot2)
library(lubridate)##
## Attaching package: 'lubridate'## The following objects are masked from 'package:base':
##
## date, intersect, setdiff, unionlibrary(janitor)##
## Attaching package: 'janitor'## The following objects are masked from 'package:stats':
##
## chisq.test, fisher.testlibrary(dplyr)2- Loading the datasets:
daily_activity <- read_csv("C:\\Users\\Lenovo\\Desktop\\Bellabeat\\Fitbit Data\\dailyActivity_merged.csv")## Rows: 940 Columns: 15
## -- Column specification --------------------------------------------------------
## Delimiter: ","
## chr (1): ActivityDate
## dbl (14): Id, TotalSteps, TotalDistance, TrackerDistance, LoggedActivitiesDi...
##
## i Use `spec()` to retrieve the full column specification for this data.
## i Specify the column types or set `show_col_types = FALSE` to quiet this message.heartrate_seconds<- read_csv("C:\\Users\\Lenovo\\Desktop\\Bellabeat\\Fitbit Data\\heartrate_seconds_merged.csv")## Rows: 2483658 Columns: 3
## -- Column specification --------------------------------------------------------
## Delimiter: ","
## chr (1): Time
## dbl (2): Id, Value
##
## i Use `spec()` to retrieve the full column specification for this data.
## i Specify the column types or set `show_col_types = FALSE` to quiet this message.sleep_day <- read_csv("C:\\Users\\Lenovo\\Desktop\\Bellabeat\\Fitbit Data\\sleepDay_merged.csv")## Rows: 413 Columns: 5
## -- Column specification --------------------------------------------------------
## Delimiter: ","
## chr (1): SleepDay
## dbl (4): Id, TotalSleepRecords, TotalMinutesAsleep, TotalTimeInBed
##
## i Use `spec()` to retrieve the full column specification for this data.
## i Specify the column types or set `show_col_types = FALSE` to quiet this message.3- Checking the structure of the datasets:
str(daily_activity)## spec_tbl_df [940 x 15] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ Id : num [1:940] 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityDate : chr [1:940] "4/12/2016" "4/13/2016" "4/14/2016" "4/15/2016" ...
## $ TotalSteps : num [1:940] 13162 10735 10460 9762 12669 ...
## $ TotalDistance : num [1:940] 8.5 6.97 6.74 6.28 8.16 ...
## $ TrackerDistance : num [1:940] 8.5 6.97 6.74 6.28 8.16 ...
## $ LoggedActivitiesDistance: num [1:940] 0 0 0 0 0 0 0 0 0 0 ...
## $ VeryActiveDistance : num [1:940] 1.88 1.57 2.44 2.14 2.71 ...
## $ ModeratelyActiveDistance: num [1:940] 0.55 0.69 0.4 1.26 0.41 ...
## $ LightActiveDistance : num [1:940] 6.06 4.71 3.91 2.83 5.04 ...
## $ SedentaryActiveDistance : num [1:940] 0 0 0 0 0 0 0 0 0 0 ...
## $ VeryActiveMinutes : num [1:940] 25 21 30 29 36 38 42 50 28 19 ...
## $ FairlyActiveMinutes : num [1:940] 13 19 11 34 10 20 16 31 12 8 ...
## $ LightlyActiveMinutes : num [1:940] 328 217 181 209 221 164 233 264 205 211 ...
## $ SedentaryMinutes : num [1:940] 728 776 1218 726 773 ...
## $ Calories : num [1:940] 1985 1797 1776 1745 1863 ...
## - attr(*, "spec")=
## .. cols(
## .. Id = col_double(),
## .. ActivityDate = col_character(),
## .. TotalSteps = col_double(),
## .. TotalDistance = col_double(),
## .. TrackerDistance = col_double(),
## .. LoggedActivitiesDistance = col_double(),
## .. VeryActiveDistance = col_double(),
## .. ModeratelyActiveDistance = col_double(),
## .. LightActiveDistance = col_double(),
## .. SedentaryActiveDistance = col_double(),
## .. VeryActiveMinutes = col_double(),
## .. FairlyActiveMinutes = col_double(),
## .. LightlyActiveMinutes = col_double(),
## .. SedentaryMinutes = col_double(),
## .. Calories = col_double()
## .. )
## - attr(*, "problems")=<externalptr>str(heartrate_seconds)## spec_tbl_df [2,483,658 x 3] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ Id : num [1:2483658] 2.02e+09 2.02e+09 2.02e+09 2.02e+09 2.02e+09 ...
## $ Time : chr [1:2483658] "4/12/2016 7:21:00 AM" "4/12/2016 7:21:05 AM" "4/12/2016 7:21:10 AM" "4/12/2016 7:21:20 AM" ...
## $ Value: num [1:2483658] 97 102 105 103 101 95 91 93 94 93 ...
## - attr(*, "spec")=
## .. cols(
## .. Id = col_double(),
## .. Time = col_character(),
## .. Value = col_double()
## .. )
## - attr(*, "problems")=<externalptr>str(sleep_day)## spec_tbl_df [413 x 5] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ Id : num [1:413] 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ SleepDay : chr [1:413] "4/12/2016 12:00:00 AM" "4/13/2016 12:00:00 AM" "4/15/2016 12:00:00 AM" "4/16/2016 12:00:00 AM" ...
## $ TotalSleepRecords : num [1:413] 1 2 1 2 1 1 1 1 1 1 ...
## $ TotalMinutesAsleep: num [1:413] 327 384 412 340 700 304 360 325 361 430 ...
## $ TotalTimeInBed : num [1:413] 346 407 442 367 712 320 377 364 384 449 ...
## - attr(*, "spec")=
## .. cols(
## .. Id = col_double(),
## .. SleepDay = col_character(),
## .. TotalSleepRecords = col_double(),
## .. TotalMinutesAsleep = col_double(),
## .. TotalTimeInBed = col_double()
## .. )
## - attr(*, "problems")=<externalptr>This shows us that:
- ActivityDate in the dataset daily_activity is a
character: needs to be converted to a Date object.
- Time in the dataset heartrate_seconds is a
character: needs to be converted to a Time object.
- SleepDay in the dataset sleep_day is a character:
needs to be converted to a DateTime object.
4- Checking for number of unique participants:
n_distinct(daily_activity$Id)## [1] 33n_distinct(heartrate_seconds$Id)## [1] 14n_distinct(sleep_day$Id)## [1] 24We notice that:
- daily_activity has 33 unique identification numbers. Since
the survey has only 30 participants, daily_activity has 3 more
participants than it should. This could be due to some people making
more than one contribution in the survey.
- heartrate_seconds and sleep_day have only 14 and 24
unique identification numbers, respectively. This means that some people
either didn’t track their heartbeat data, or the number of hours that
they’ve spent asleep per day, or both.
5- Checking for duplicates:
sum(duplicated(daily_activity))## [1] 0sum(duplicated(heartrate_seconds))## [1] 0sum(duplicated(sleep_day))## [1] 3- sleep_day has 3 duplicated rows. I will be using the function distinct() from the dplyr package (already loaded) to remove the duplicates.
sleep_day <- sleep_day %>% distinct()6- Checking for missing values:
sum(is.na(daily_activity))## [1] 0sum(is.na(heartrate_seconds))## [1] 0sum(is.na(sleep_day))## [1] 0colSums(daily_activity==0)## Id ActivityDate TotalSteps
## 0 0 77
## TotalDistance TrackerDistance LoggedActivitiesDistance
## 78 78 908
## VeryActiveDistance ModeratelyActiveDistance LightActiveDistance
## 413 386 85
## SedentaryActiveDistance VeryActiveMinutes FairlyActiveMinutes
## 858 409 384
## LightlyActiveMinutes SedentaryMinutes Calories
## 84 1 4colSums(heartrate_seconds==0)## Id Time Value
## 0 0 0colSums(sleep_day==0)## Id SleepDay TotalSleepRecords TotalMinutesAsleep
## 0 0 0 0
## TotalTimeInBed
## 0- We can see that daily_activity has values that are equal to 0 in multiple columns. (This could be due to some people having made no exercise or walks on some days, or failing to wear their smart devices when exercising,…)
7- Checking the data for errors:
- The column names in all the datasets need to be changed, to be made more consistent. I will be using the function clean_names() from the janitor package (already loaded).
daily_activity <- clean_names(daily_activity)
heartrate_seconds <- clean_names(heartrate_seconds)
sleep_day <- clean_names(sleep_day)a) daily_activity:
daily_activity %>% mutate(total_distance_calculated=very_active_distance+moderately_active_distance+light_active_distance+sedentary_active_distance) %>% select(c(total_distance,total_distance_calculated))## # A tibble: 940 x 2
## total_distance total_distance_calculated
## <dbl> <dbl>
## 1 8.5 8.49
## 2 6.97 6.97
## 3 6.74 6.75
## 4 6.28 6.23
## 5 8.16 8.16
## 6 6.48 6.48
## 7 8.59 8.60
## 8 9.88 9.88
## 9 6.68 6.68
## 10 6.34 6.34
## # ... with 930 more rowsRunning the code above shows that some values between the total_distance_calculated (now), and the total_distance already given in the dataset, slightly differ. However, this error is small and doesn’t affect the analysis results.
Now, the column ActivityDate needs to be converted from Character to Date, as stated before. I’ll be using the lubridate package (already loaded) to make that conversion.
daily_activity$activity_date <- mdy(daily_activity$activity_date)- Since the survey was collected over a 2 month period (from 2016-04-12 to 2016-05-12), I’m gonna be checking that the dates in daily_activity all fall in the right interval.
sum(daily_activity$activity_date<"2016-04-12"|daily_activity$activity_date>"2016-05-12")## [1] 0- After adding new columns that show the number of hours people spent on different activity levels (very_active_hours, fairly_active_hours,lightly_active_hours, sedentary_hours), and the total number of hours that they spent active (total_active_hours), I will remove the columns that used to show this data in minutes.
daily_activity <- daily_activity %>% mutate(very_active_hours=very_active_minutes/60,fairly_active_hours=fairly_active_minutes/60,lightly_active_hours=lightly_active_minutes/60,sedentary_hours=sedentary_minutes/60)
daily_activity <- daily_activity %>% mutate(total_active_hours=very_active_hours+fairly_active_hours+lightly_active_hours+sedentary_hours)
daily_activity <- daily_activity %>% select(-c(very_active_minutes,fairly_active_minutes,lightly_active_minutes,sedentary_minutes))b) heartrate_seconds:
First, (to conserve the AM/PM data), before separating the time column, it needs to be converted from Character to a DateTime object. I will be using the lubridate package (already loaded).
Then, using the format() function, I will omit all seconds from the given time (since a) they add no important value to the data, and b) to later be able to calculate the average heartbeat per minute, and then group together the average heartbeat per day, for each id).
heartrate_seconds$time <- mdy_hms(heartrate_seconds$time)
heartrate_seconds$time <- format(heartrate_seconds$time,'%Y-%m-%d %H:%M')
heartrate_seconds <- heartrate_seconds %>% separate(time,c("date","time"),' ')
str(heartrate_seconds)## tibble [2,483,658 x 4] (S3: tbl_df/tbl/data.frame)
## $ id : num [1:2483658] 2.02e+09 2.02e+09 2.02e+09 2.02e+09 2.02e+09 ...
## $ date : chr [1:2483658] "2016-04-12" "2016-04-12" "2016-04-12" "2016-04-12" ...
## $ time : chr [1:2483658] "07:21" "07:21" "07:21" "07:21" ...
## $ value: num [1:2483658] 97 102 105 103 101 95 91 93 94 93 ...- After separating the original time column into new separate date and time columns and checking with str(), we notice that the date and time reverted back to Character objects, and need to be converted again to Date and Time objects, respectively (using the lubridate package).
heartrate_seconds$date <- ymd(heartrate_seconds$date)
heartrate_seconds$time <- hm(heartrate_seconds$time)- Now, (as done for daily_activity earlier), I will again check that the given dates are included in the required 2 month-interval.
sum(heartrate_seconds$date<"2016-04-12"||heartrate_seconds$date>"2016-05-12")## [1] 0- I will now create a new table (heartrate) with the average heartbeat recorded per day for each individual.
heartrate <- heartrate_seconds %>% group_by(id,date,time) %>% drop_na() %>% summarize(average_heartbeat = mean(value))## `summarise()` has grouped output by 'id', 'date'. You can override using the
## `.groups` argument.heartrate <- heartrate %>% select(-time)
heartrate <- heartrate %>% group_by(id,date) %>% drop_na() %>% summarise(heartbeat=mean(average_heartbeat))## `summarise()` has grouped output by 'id'. You can override using the `.groups`
## argument.heartrate$heartbeat <- as.integer(heartrate$heartbeat)
head(heartrate)## # A tibble: 6 x 3
## # Groups: id [1]
## id date heartbeat
## <dbl> <date> <int>
## 1 2022484408 2016-04-12 74
## 2 2022484408 2016-04-13 78
## 3 2022484408 2016-04-14 70
## 4 2022484408 2016-04-15 78
## 5 2022484408 2016-04-16 74
## 6 2022484408 2016-04-17 82c) sleep_day:
I will split the sleep_day column into 2 separate columns: sleep_date and sleep_time.
Checking with str() shows us that sleep_date needs to be changed from Character to Date, and sleep_time from Character to Time (using the lubridate package).
sleep_day <- sleep_day %>% separate(sleep_day,c("sleep_date","sleep_time"),sep=' ')## Warning: Expected 2 pieces. Additional pieces discarded in 410 rows [1, 2, 3, 4,
## 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, ...].sleep_day$sleep_date <- mdy(sleep_day$sleep_date)
sleep_day$sleep_time <- hms(sleep_day$sleep_time)- I will convert the data to be shown in hours, and then remove the data that was given in minutes:
sleep_day <- mutate(sleep_day,total_hours_asleep=total_minutes_asleep/60)
sleep_day <- mutate(sleep_day,total_hours_in_bed=total_time_in_bed/60)
sleep_day <- sleep_day %>% select(-c(total_minutes_asleep,total_time_in_bed))- Checking for errors:
- if total_hours has any data that is negative or exceeds 24 hours:
sum(sleep_day$total_hours_asleep>=24|sleep_day$total_hours_asleep<=0,na.rm=TRUE)## [1] 0sum(sleep_day$total_hours_in_bed>=24||sleep_day$total_hours_in_bed<=0,na.rm=TRUE)## [1] 0- if sleep_date is consistent with the given date interval:
sum(sleep_day$sleep_date<"2016-04-12"||sleep_day$sleep_date>"2016-05-12")## [1] 0- if any total hours spent asleep exceeds the total time spent in bed:
sum(sleep_day$total_hours_asleep>sleep_day$total_hours_in_bed,na.rm=TRUE)## [1] 04- Analyze:
- Now, to be able to analyze the data, I’m first gonna use left_join() to join the daily_activity and heartrate tables. Then, using the same function again, I’m gonna join the result with the sleep_day table.
join_table <- left_join(daily_activity,heartrate,by=c("id","activity_date"="date"))
join_table <- left_join(join_table,sleep_day,by=c("id","activity_date"="sleep_date"))- I’m gonna create the final table to use for the analysis (final_table), where I will remove some unnecessary data from the joined tables earlier. Then I’m gonna use summary() to do some statistical calculations on the data.
final_table <- select(join_table,-c(tracker_distance,logged_activities_distance,sleep_time))
final_table %>% select(-c(id,activity_date)) %>% drop_na() %>% summary()## total_steps total_distance very_active_distance
## Min. : 42 Min. : 0.030 Min. :0.000
## 1st Qu.: 5245 1st Qu.: 3.680 1st Qu.:0.000
## Median : 9148 Median : 6.380 Median :0.670
## Mean : 8522 Mean : 6.021 Mean :1.429
## 3rd Qu.:11135 3rd Qu.: 7.920 3rd Qu.:2.420
## Max. :22770 Max. :17.540 Max. :9.890
## moderately_active_distance light_active_distance sedentary_active_distance
## Min. :0.0000 Min. :0.030 Min. :0.000000
## 1st Qu.:0.0000 1st Qu.:2.670 1st Qu.:0.000000
## Median :0.4000 Median :3.770 Median :0.000000
## Mean :0.6475 Mean :3.894 Mean :0.002099
## 3rd Qu.:1.0000 3rd Qu.:5.270 3rd Qu.:0.000000
## Max. :2.7700 Max. :9.480 Max. :0.110000
## calories very_active_hours fairly_active_hours lightly_active_hours
## Min. : 403 Min. :0.0000 Min. :0.0000 Min. :0.06667
## 1st Qu.:1994 1st Qu.:0.0000 1st Qu.:0.0000 1st Qu.:2.93333
## Median :2225 Median :0.2000 Median :0.2167 Median :3.78333
## Mean :2467 Mean :0.4666 Mean :0.2766 Mean :3.81077
## 3rd Qu.:3004 3rd Qu.:0.6333 3rd Qu.:0.4833 3rd Qu.:4.61667
## Max. :4900 Max. :3.5000 Max. :1.2333 Max. :8.63333
## sedentary_hours total_active_hours heartbeat total_sleep_records
## Min. : 0.03333 Min. : 0.10 Min. : 58.0 Min. :1.000
## 1st Qu.:10.35000 1st Qu.:15.35 1st Qu.: 65.0 1st Qu.:1.000
## Median :11.78333 Median :16.42 Median : 70.0 Median :1.000
## Mean :11.69954 Mean :16.25 Mean : 70.9 Mean :1.122
## 3rd Qu.:13.01667 3rd Qu.:17.37 3rd Qu.: 75.0 3rd Qu.:1.000
## Max. :20.36667 Max. :23.30 Max. :100.0 Max. :3.000
## total_hours_asleep total_hours_in_bed
## Min. : 0.9667 Min. : 1.017
## 1st Qu.: 6.5167 1st Qu.: 6.883
## Median : 7.3167 Median : 7.733
## Mean : 7.1041 Mean : 7.580
## 3rd Qu.: 8.2000 3rd Qu.: 8.767
## Max. :12.9167 Max. :14.050- We can see that the average total number of steps per day and per person is 8522. This is less than the recommended 10 000 steps/day by the CDC.
- We can see that each person spent an average of 16.25 total active hours per day. However, on average, only half an hour approximately (3%) was spent being very active, while roughly 12 hours were spent being sedentary (roughly 72%).
- We can see that the average heart rate recorded per person is 79 beats per minute. This falls in the normal resting heart rate range for adults in general (between 60 and 100 beats per minute).
- On average, each person has slept once everyday, with approximately 7.5 hours spent in bed, and 7 hours spent asleep. This shows that the participants only spent time in bed when needing to sleep.
5- Share:
final_table %>% drop_na() %>% ggplot(aes(x=sedentary_hours,y=total_hours_asleep)) + geom_point(color="red")+geom_smooth()+labs(title="The relationship between Sedentary Hours and Total Hours Asleep",x="Sedentary Hours",y="Total Hours Asleep")## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
- We can see a negative linear relationship between sedentary hours and the total hours spent asleep. This shows that the less exercise (or even simple walking) people made, the less sleep they had. This can mean that an increase in physical activity may lead to better sleep quality.
final_table %>% drop_na() %>% ggplot(aes(x=total_steps,y=calories)) + geom_point(color="blue")+geom_smooth()+labs(title="The relationship between Total Steps and burned Calories",x="Total Steps",y="calories")## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
- We can see a positive correlation between the total steps taken per day, and the number of calories burned. This shows that the more people move around and exercise, the more calories they are able to burn.
final_table %>% drop_na() %>% ggplot(aes(x=total_hours_in_bed,y=total_hours_asleep)) + geom_point(color="purple")+geom_smooth()+labs(title="The relationship between Total Hours In Bed and Total Hours Asleep",x="Total Hours In Bed",y="Total Hours Asleep")## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
- We clearly see a positive correlation between the total hours spent in bed and the total hours spent asleep.
final_table %>% drop_na() %>% ggplot(aes(x=total_distance,y=total_steps)) + geom_point(color="green")+geom_smooth()+labs(title="The relationship between Total Distance and Total Steps",x="Total Distance",y="Total Steps")## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
- We can see a clear positive correlation between the total active distance per day, and the total number of steps taken per day. Meaning the more exercise people made, the more distance they walked.
5- Act:
- Recommendations:
- Since the number of steps taken per day is less than the CDC
recommended 10 000 steps (this could be due to people getting busy at
work and forgetting to move around), Bellabeat should send notifications
every hour or so on the chosen smart device to remind their customers to
get up and walk for a minimum of 10 minutes.
- Since the total number of very active hours per day is low, Bellabeat should create a program where people earn money, or coins they can use to buy premium products, whenever they hit the recommended hours and intensity of activity. This could encourage people to exercise more, which then, as we saw in our analysis, will help them burn more calories, and will improve their quality and time of sleep.
I’m self-learning data analysis and this is my first time using R or trying to clean/analyze data. If there are any notes on my code or mistakes in it, please do not hesitate to let me know. I would love feedback on how I can improve. Thank you for anyone who checked out my case study.