https://github.com/anthonysanalysis/bellabeat-analysis
Bellabeat Tech Case Study Capstone Project
https://github.com/anthonysanalysis/bellabeat-analysis
analysis capstone case-study data data-analysis data-visualization md r rmd rstudio
Last synced: about 1 month ago
JSON representation
Bellabeat Tech Case Study Capstone Project
- Host: GitHub
- URL: https://github.com/anthonysanalysis/bellabeat-analysis
- Owner: AnthonysAnalysis
- Created: 2025-01-25T00:22:15.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-01-28T00:26:52.000Z (over 1 year ago)
- Last Synced: 2025-03-28T15:15:12.450Z (about 1 year ago)
- Topics: analysis, capstone, case-study, data, data-analysis, data-visualization, md, r, rmd, rstudio
- Homepage:
- Size: 1.95 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Bellabeat Tech Capstone Project
#### Author: Anthony Tran
#### Date: 2024-12-20
# **About:**
Bellabeat is a small, successful manufacturer of health-focused tech
products for women. They are looking to expand their market share in the
global smart-device market.
I, as a junior data analyst on the marketing analyst team have been
tasked with working on one of Bellabeat’s products, their app which
connects to their many smart devices. We will be analyzing usage data on
one of Bellabeat’s competitor’s devices, the Fitbit, to gain insight
into how customers are using their smart devices, which will help create
data-driven decision making for the company’s marketing strategy for
growth.
Throughout the project, we will follow the six main steps of the data
analysis process:
**Ask, prepare, process, analyze, share, and act.**
# **Ask -**
Business task:
We will identify usage and data trends in the Fitbit, a handheld
wearable from one of Bellabeat’s competitors. We will run some data
analysis and acquire some insight into this smart device’s usage. Then,
figure out how to apply these insights into our own products to drive
Bellabeat’s marketing strategy via data-driven decision making.
# **Prepare -**
Datasource: Our dataset is the Fitbit Fitness Tracker Dataset (CC0:
Public Domain), via Mobius.
Accessibility and privacy: The data is open-sourced and licensed under
CC0: Public Domain in which the original owner relinquished all
copyright and similar rights worldwide, and dedicated those rights to
the public domain. Others can copy, modify, distribute and perform the
work, even for commercial purposes, all without asking permission.
Dataset information: This dataset contains data generated from ~ 33
Fitbit users respondents to an Amazon Mechanical Turk survey, who
consented to the submission of their personal tracker data from the
timeframe 04/12/2016 - 05/12/2016. Various different quantitative data
metrics include physical activity, heart rate, and sleep monitoring at
different levels of scope, including minute-output,hourly, and daily.
Variation between output represents use of different types of Fitbit
trackers and individual tracking behaviors / preferences.
Data organization: The data was made available via 18 .csv files, each
containing various different quantitative data metrics.The majority of
which is in long format, with 3 files in wide format. Some datasets were
excluded from analysis after verification due to being redundant data
already included in a more complete data table, or the sample size for
participant response was too small to generate significant insight.
Data limitations: Unknown demographics. We do not know various
demographics for our participants such as age or sex, which would be
important especially as it relates to our company which focuses on
women-centric products. Also, some of the data given are not clearly
labeled with the proper units, so we had to assume some standard units
to our best guess, given the context.
Small sample size and time frame: In statistics, the golden rule for
analysis is that carries the most power when the sample size is over 30,
the larger the better. For most of our data, we are barely at the bare
minimum so insight might not be able to be extrapolated strongly to a
larger population or timeframe.
# **Process -**
Analysis will be carried out in programming language R, in Posit Cloud
due to ease of use for statistical computing, data visualization, and
sharing results with stakeholders.
## Setting up my environment
Notes: We’ll begin by installing and loading the necessary packages to
run our analysis:
tidyverse, dplyr, here, janitor, skimr, lubridate, tibble, and waffle.
``` r
install.packages("tidyverse")
install.packages("dplyr")
install.packages("here")
install.packages("janitor")
install.packages("skimr")
install.packages("lubridate")
install.packages("tibble")
install.packages("waffle")
library(tidyverse)
library(dplyr)
library(here)
library(janitor)
library(skimr)
library(lubridate)
library(tibble)
library(waffle)
```
## Importing our datasets
Notes: We need to import our various datasets, which are in the form of
a .csv, into data frames that we can work with into R.
``` r
daily_activity <- read.csv("dailyActivity_merged.csv", header = TRUE)
daily_calories <- read.csv("dailyCalories_merged.csv", header = TRUE)
daily_intensities <- read.csv("dailyIntensities_merged.csv", header = TRUE)
daily_steps <- read.csv("dailySteps_merged.csv", header = TRUE)
sleep_day <- read.csv("sleepDay_merged.csv", header = TRUE)
weight_log <- read.csv("weightLogInfo_merged.csv", header = TRUE)
hourly_steps <- read.csv("hourlySteps_merged.csv", header = TRUE)
hourly_intensity <- read.csv("hourlyIntensities_merged.csv", header = TRUE)
hourly_calories <- read.csv("hourlyCalories_merged.csv", header = TRUE)
```
## Previewing datasets
Notes: Getting a preview and examining the structure of our recently
imported datasets for preliminary analysis and planning.
``` r
str(daily_activity)
```
## 'data.frame': 940 obs. of 15 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityDate : chr "4/12/2016" "4/13/2016" "4/14/2016" "4/15/2016" ...
## $ TotalSteps : int 13162 10735 10460 9762 12669 9705 13019 15506 10544 9819 ...
## $ TotalDistance : num 8.5 6.97 6.74 6.28 8.16 ...
## $ TrackerDistance : num 8.5 6.97 6.74 6.28 8.16 ...
## $ LoggedActivitiesDistance: num 0 0 0 0 0 0 0 0 0 0 ...
## $ VeryActiveDistance : num 1.88 1.57 2.44 2.14 2.71 ...
## $ ModeratelyActiveDistance: num 0.55 0.69 0.4 1.26 0.41 ...
## $ LightActiveDistance : num 6.06 4.71 3.91 2.83 5.04 ...
## $ SedentaryActiveDistance : num 0 0 0 0 0 0 0 0 0 0 ...
## $ VeryActiveMinutes : int 25 21 30 29 36 38 42 50 28 19 ...
## $ FairlyActiveMinutes : int 13 19 11 34 10 20 16 31 12 8 ...
## $ LightlyActiveMinutes : int 328 217 181 209 221 164 233 264 205 211 ...
## $ SedentaryMinutes : int 728 776 1218 726 773 539 1149 775 818 838 ...
## $ Calories : int 1985 1797 1776 1745 1863 1728 1921 2035 1786 1775 ...
``` r
str(daily_calories)
```
## 'data.frame': 940 obs. of 3 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityDay: chr "4/12/2016" "4/13/2016" "4/14/2016" "4/15/2016" ...
## $ Calories : int 1985 1797 1776 1745 1863 1728 1921 2035 1786 1775 ...
``` r
str(daily_intensities)
```
## 'data.frame': 940 obs. of 10 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityDay : chr "4/12/2016" "4/13/2016" "4/14/2016" "4/15/2016" ...
## $ SedentaryMinutes : int 728 776 1218 726 773 539 1149 775 818 838 ...
## $ LightlyActiveMinutes : int 328 217 181 209 221 164 233 264 205 211 ...
## $ FairlyActiveMinutes : int 13 19 11 34 10 20 16 31 12 8 ...
## $ VeryActiveMinutes : int 25 21 30 29 36 38 42 50 28 19 ...
## $ SedentaryActiveDistance : num 0 0 0 0 0 0 0 0 0 0 ...
## $ LightActiveDistance : num 6.06 4.71 3.91 2.83 5.04 ...
## $ ModeratelyActiveDistance: num 0.55 0.69 0.4 1.26 0.41 ...
## $ VeryActiveDistance : num 1.88 1.57 2.44 2.14 2.71 ...
``` r
str(daily_steps)
```
## 'data.frame': 940 obs. of 3 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityDay: chr "4/12/2016" "4/13/2016" "4/14/2016" "4/15/2016" ...
## $ StepTotal : int 13162 10735 10460 9762 12669 9705 13019 15506 10544 9819 ...
``` r
str(sleep_day)
```
## 'data.frame': 413 obs. of 5 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ SleepDay : chr "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 : int 1 2 1 2 1 1 1 1 1 1 ...
## $ TotalMinutesAsleep: int 327 384 412 340 700 304 360 325 361 430 ...
## $ TotalTimeInBed : int 346 407 442 367 712 320 377 364 384 449 ...
``` r
str(weight_log)
```
## 'data.frame': 67 obs. of 8 variables:
## $ Id : num 1.50e+09 1.50e+09 1.93e+09 2.87e+09 2.87e+09 ...
## $ Date : chr "5/2/2016 11:59:59 PM" "5/3/2016 11:59:59 PM" "4/13/2016 1:08:52 AM" "4/21/2016 11:59:59 PM" ...
## $ WeightKg : num 52.6 52.6 133.5 56.7 57.3 ...
## $ WeightPounds : num 116 116 294 125 126 ...
## $ Fat : int 22 NA NA NA NA 25 NA NA NA NA ...
## $ BMI : num 22.6 22.6 47.5 21.5 21.7 ...
## $ IsManualReport: chr "True" "True" "False" "True" ...
## $ LogId : num 1.46e+12 1.46e+12 1.46e+12 1.46e+12 1.46e+12 ...
``` r
str(hourly_steps)
```
## 'data.frame': 22099 obs. of 3 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityHour: chr "4/12/2016 12:00:00 AM" "4/12/2016 1:00:00 AM" "4/12/2016 2:00:00 AM" "4/12/2016 3:00:00 AM" ...
## $ StepTotal : int 373 160 151 0 0 0 0 0 250 1864 ...
``` r
str(hourly_intensity)
```
## 'data.frame': 22099 obs. of 4 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityHour : chr "4/12/2016 12:00:00 AM" "4/12/2016 1:00:00 AM" "4/12/2016 2:00:00 AM" "4/12/2016 3:00:00 AM" ...
## $ TotalIntensity : int 20 8 7 0 0 0 0 0 13 30 ...
## $ AverageIntensity: num 0.333 0.133 0.117 0 0 ...
``` r
str(hourly_calories)
```
## 'data.frame': 22099 obs. of 3 variables:
## $ Id : num 1.5e+09 1.5e+09 1.5e+09 1.5e+09 1.5e+09 ...
## $ ActivityHour: chr "4/12/2016 12:00:00 AM" "4/12/2016 1:00:00 AM" "4/12/2016 2:00:00 AM" "4/12/2016 3:00:00 AM" ...
## $ Calories : int 81 61 59 47 48 48 48 47 68 141 ...
Upon examining the data frames and column names, we see that
daily_activities also contains the complete data set for daily_calories,
daily_intensities, and daily_steps as well.
So from now on when using the data for daily_calories,
daily_intensities, and daily_steps, I will reference their respective
columns from the daily_activity data frame for simplicity.
## Verifying sample size
Notes: Counting how many distinct participants we have for each dataset,
and whether it is appropriate to use for our analysis.
``` r
n_distinct(daily_activity$Id)
```
## [1] 33
``` r
n_distinct(daily_calories$Id)
```
## [1] 33
``` r
n_distinct(daily_intensities$Id)
```
## [1] 33
``` r
n_distinct(daily_steps$Id)
```
## [1] 33
``` r
n_distinct(sleep_day$Id)
```
## [1] 24
``` r
n_distinct(weight_log$Id)
```
## [1] 8
``` r
n_distinct(hourly_steps$Id)
```
## [1] 33
``` r
n_distinct(hourly_intensity$Id)
```
## [1] 33
``` r
n_distinct(hourly_calories$Id)
```
## [1] 33
In statistics, there is a “rule of 30,” where 30 is typically the
minimum sample size needed to make a statistically strong conclusion
about the data. So because of this I will discard the weight data, which
only has 8 people. Sleep has 24 participants, but I will still run a
preliminary analysis on it regardless, with that in consideration.
## Converting into proper date format
We also see from the previous section that the dates are in character
format, so we will need to convert to proper date or date/time format,
and hide the old columns.
This is so our packages will be able to properly run on the data in the
column.
``` r
daily_activity <- daily_activity %>%
mutate(ActivityDate = mdy(ActivityDate))
hourly_steps <- hourly_steps %>%
mutate(ActivityDateHour = mdy_hms(ActivityHour)) %>%
select(-ActivityHour)
hourly_intensity <- hourly_intensity %>%
mutate(ActivityDateHour = mdy_hms(ActivityHour)) %>%
select(-ActivityHour)
hourly_calories <- hourly_calories %>%
mutate(ActivityDateHour = mdy_hms(ActivityHour)) %>%
select(-ActivityHour)
sleep_day <- sleep_day %>%
mutate(SleepDate = as.Date(mdy_hms(SleepDay))) %>%
select(-SleepDay)
```
## Verifying the values were properly changed from character to date & date/time.
``` r
class(daily_activity$ActivityDate)
```
## [1] "Date"
``` r
class(hourly_steps$ActivityDateHour)
```
## [1] "POSIXct" "POSIXt"
``` r
class(hourly_intensity$ActivityDateHour)
```
## [1] "POSIXct" "POSIXt"
``` r
class(hourly_calories$ActivityDateHour)
```
## [1] "POSIXct" "POSIXt"
``` r
class(sleep_day$SleepDate)
```
## [1] "Date"
## Cleaning the data
Notes: Each data frame was manually imported and assigned the
appropriate naming convention by me in snakecase.
Next we will check for errors and inconsistencies in the data by
checking for duplicated records and missing values.
``` r
sum(duplicated(daily_activity))
```
## [1] 0
``` r
sum(duplicated(sleep_day))
```
## [1] 3
``` r
sum(duplicated(hourly_steps))
```
## [1] 0
``` r
sum(duplicated(hourly_calories))
```
## [1] 0
``` r
sum(duplicated(hourly_intensity))
```
## [1] 0
``` r
sum(is.na(daily_activity))
```
## [1] 0
``` r
sum(is.na(sleep_day))
```
## [1] 0
``` r
sum(is.na(hourly_steps))
```
## [1] 0
``` r
sum(is.na(hourly_calories))
```
## [1] 0
``` r
sum(is.na(hourly_intensity))
```
## [1] 0
sleep_day has 3 duplicated records. All other datasets passed the
initial check.
## Get rid of duplicated observations, and drop missing values from data frames as needed.
``` r
daily_activity <- daily_activity %>%
distinct %>%
drop_na
sleep_day <-sleep_day %>%
distinct %>%
drop_na
hourly_steps <- hourly_steps %>%
distinct %>%
drop_na
hourly_calories <- hourly_calories %>%
distinct %>%
drop_na
hourly_intensity <- hourly_intensity %>%
distinct %>%
drop_na
```
## Merge to create a new combined data frame, and verify distinct users.
``` r
daily_activity_sleep <- merge(daily_activity, sleep_day, by.x = c("Id", "ActivityDate"), by.y = c("Id", "SleepDate"), all = TRUE)
n_distinct(daily_activity_sleep$Id)
```
## [1] 33
Here we are creating a newly merged data frame for one of our analyses
that we are going to run.
# **Analyze & Share -**
## Basic overview summary:
Notes: Some basic preliminary summary analysis
``` r
daily_activity %>%
select(TotalSteps, Calories, SedentaryMinutes) %>%
summary()
```
## TotalSteps Calories SedentaryMinutes
## Min. : 0 Min. : 0 Min. : 0.0
## 1st Qu.: 3790 1st Qu.:1828 1st Qu.: 729.8
## Median : 7406 Median :2134 Median :1057.5
## Mean : 7638 Mean :2304 Mean : 991.2
## 3rd Qu.:10727 3rd Qu.:2793 3rd Qu.:1229.5
## Max. :36019 Max. :4900 Max. :1440.0
``` r
sleep_day %>%
select(TotalSleepRecords, TotalMinutesAsleep, TotalTimeInBed) %>%
summary()
```
## TotalSleepRecords TotalMinutesAsleep TotalTimeInBed
## Min. :1.00 Min. : 58.0 Min. : 61.0
## 1st Qu.:1.00 1st Qu.:361.0 1st Qu.:403.8
## Median :1.00 Median :432.5 Median :463.0
## Mean :1.12 Mean :419.2 Mean :458.5
## 3rd Qu.:1.00 3rd Qu.:490.0 3rd Qu.:526.0
## Max. :3.00 Max. :796.0 Max. :961.0
``` r
hourly_steps %>%
select(StepTotal) %>%
filter(StepTotal != 0) %>%
summary()
```
## StepTotal
## Min. : 1.0
## 1st Qu.: 103.0
## Median : 287.0
## Mean : 552.7
## 3rd Qu.: 643.0
## Max. :10554.0
``` r
hourly_calories %>%
select(Calories) %>%
summary()
```
## Calories
## Min. : 42.00
## 1st Qu.: 63.00
## Median : 83.00
## Mean : 97.39
## 3rd Qu.:108.00
## Max. :948.00
``` r
hourly_intensity %>%
select(TotalIntensity, AverageIntensity) %>%
summary()
```
## TotalIntensity AverageIntensity
## Min. : 0.00 Min. :0.0000
## 1st Qu.: 0.00 1st Qu.:0.0000
## Median : 3.00 Median :0.0500
## Mean : 12.04 Mean :0.2006
## 3rd Qu.: 16.00 3rd Qu.:0.2667
## Max. :180.00 Max. :3.0000
- The Center for Disease Control (CDC) recommends that the average
adult take around 8,000-10,000+ steps per day. Compared to our data
showing the mean and median daily steps at 7638 and 7406,
respectively. Adults are not being as active as they should be. One
premature recommendation is a device feature reminding people to get
up and be active/walk around once in a while, after a certain amount
of sedentary minutes has elapsed.
- The National Institute of Health (NIH) recommends that adults get
between 7 and 9 hours of sleep per night. The average sleep time for
our group here is just around 7 hours, barely at the lower end of
the minimum. The data shows that this sample does not meet the
recommended goal. We can use this as an opportunity to add a feature
to our device that reminds users to go to bed at a reasonable time,
given what time they wake up. As well as add in these
recommendations from government institutions as a credible source to
help persuade people to comply.
- The average number of hourly steps during waking hours for the
participants in our study is 552.7 steps, with a median of 287.0.
This indicates the data is skewed to the right.
- The average number of hourly calories burned for our study
participants is ~98 calories.
- The median hourly total intensity for the FitBit users is 3.00, with
a mean of 12.04, also indicating the data is skewed right.
## Let days where users used at least one functionality of the device count as an “active day.”
Out of a possible 31 days.
Let us classify usage days as:
- 0-10 Low usage
- 11-20 Medium usage
- 21-31 High usage
``` r
usage_counts <- daily_activity_sleep %>%
group_by(Id) %>%
summarise(days_active = n_distinct(ActivityDate)) %>%
mutate(device_usage = case_when(
days_active >= 0 & days_active <= 10 ~ "Low Usage",
days_active >= 11 & days_active <= 20 ~ "Medium Usage",
days_active >= 21 & days_active <= 31 ~ "High Usage"
)) %>%
group_by(device_usage) %>%
count(device_usage)
print(usage_counts)
```
## # A tibble: 3 × 2
## # Groups: device_usage [3]
## device_usage n
##
## 1 High Usage 29
## 2 Low Usage 1
## 3 Medium Usage 3
``` r
waffle_data <- setNames(usage_counts$n, usage_counts$device_usage)
waffle(waffle_data,
rows = 3,
xlab = "1 Square = 1 User",
title = "Device Usage Frequency by Total Days",
colors = c("green", "red", "yellow"),
size = 1,
legend_pos = "right")
```
- The majority of our sample participants have engaged with their
device at least once per day throughout our study period, most
fitting into our criteria of “high usage,” which is at least over
2/3rds of the total time frame of 31 days.
- We can market our more premium subscription tiers for the high usage
class who are clearly into their health metrics. We can also do this
with different price plans and types to capture the different ways
people prefer to pay, as well as for those hesitant. Possibly one
that offers more metrics, recommendations, and reminders based on
different tiers.
- Can we determine how we can get the low/medium usage groups to
engage more often?
- What are the reasons behind the low usage group and is it worth
looking into?
## Visualizations
What else can we interpret from creating some visuals?
``` r
calories_distance_model <- lm(TotalDistance ~ Calories, data = na.omit(daily_activity))
calories_distance_r_squared <- summary(calories_distance_model)$r.squared
ggplot(data = na.omit(daily_activity), aes(x = Calories, y = TotalDistance)) +
geom_point(color = "blue") +
geom_smooth(method = "lm", color = "yellow") +
labs(title = "Longer Distances Walked Leads to More Calories Burned",
x = "Daily Calories Burned",
y = "Total Daily Distance Walked") +
theme_minimal() +
annotate("text", x = max(na.omit(daily_activity)$Calories) * 0.7,
y = max(na.omit(daily_activity)$TotalDistance) * 0.9,
label = paste("R² = ", round(calories_distance_r_squared, 3)), size = 5, color = "black")
```
## `geom_smooth()` using formula = 'y ~ x'
- The general trend here is that there is a positive correlation
between total daily distance walked and daily calories burned. We
can include this information in an pop-up reminder for the user to
be aware of, if that’s their goal.
``` r
sleep_day_of_week <- sleep_day %>%
mutate(day_of_week = wday(SleepDate, label = TRUE, abbr = FALSE))
average_sleep_per_day <- sleep_day_of_week %>%
group_by(day_of_week) %>%
summarise(average_sleep_time = mean(TotalMinutesAsleep, na.rm = TRUE)) %>%
arrange(match(day_of_week, c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday")))
overall_mean_sleep <- mean(average_sleep_per_day$average_sleep_time)
ggplot(average_sleep_per_day, aes(x = day_of_week, y = average_sleep_time, fill = day_of_week)) +
geom_col() +
geom_hline(yintercept = overall_mean_sleep, linetype = "dashed", color = "red") +
labs( title = "Why Do People Sleep More on Wednesdays?", x = "Day of the Week", y = "Average Time Asleep (minutes)") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
```
- It looks like people sleep slightly more on the weekend on average.
This is expected as people are recovering from the workweek and tend
to sleep in.
- Possible point of future analysis is why it appears people on
average also sleep more on Wednesdays as well.
``` r
hourly_calories_by_hour <- hourly_calories %>%
mutate(hour_of_day = hour(ActivityDateHour))
average_hourly_calories <- hourly_calories_by_hour %>%
group_by(hour_of_day) %>%
summarise(average_calories = mean(Calories, na.rm = TRUE)) %>%
arrange(hour_of_day)
ggplot(average_hourly_calories, aes(x = hour_of_day, y = average_calories, fill = factor(hour_of_day))) +
geom_bar(stat = "identity", show.legend = FALSE) +
labs(title = "Most Hourly Calories Are Burned During Active Hours",
x = "Hour of Day",
y = "Average Calories Burned" ) +
theme_minimal() +
scale_x_continuous(breaks = 0:23) +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
```
- The most active hours of the day are during the waking hours.
- Most people wake up around 6-7am for the day, with the calories
burned gradually ramping up hourly until they get off of work at
around 4-5pm. Then presumably they would go take care of other
miscellaneous things for a little while, before hourly calories
burned comes back down towards the end of the day.
``` r
hourly_intensity_by_hour <- hourly_intensity %>%
mutate( hour_of_day = hour(ActivityDateHour),
day_type = if_else(wday(ActivityDateHour, label = TRUE) %in% c("Sat", "Sun"), "Weekend", "Weekday"))
average_hourly_intensity_by_weekend_weekday <- hourly_intensity_by_hour %>%
group_by(day_type, hour_of_day) %>% summarise(mean_intensity_hour = mean(TotalIntensity, na.rm = TRUE)) %>%
arrange(factor(day_type, levels = c("Weekday", "Weekend")), hour_of_day)
```
## `summarise()` has grouped output by 'day_type'. You can override using the
## `.groups` argument.
``` r
ggplot(average_hourly_intensity_by_weekend_weekday, aes(x = hour_of_day, y = mean_intensity_hour, fill = day_type)) +
geom_bar(stat = "identity", position = "dodge") +
labs( title = "People's Routines Differ on Weekdays vs Weekends",
x = "Hour of Day",
y = "Mean Intensity Value" ) +
scale_x_continuous(breaks = 0:23) +
scale_fill_manual(values = c("Weekday" = "blue", "Weekend" = "pink")) +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
```
- People have different schedules/routines on the weekend vs weekday.
- We can take this into consideration when designing our app features.
``` r
distance_calories_model <- lm(Calories ~ TotalDistance, data = daily_activity)
distance_calories_r_squared <- summary(distance_calories_model)$r.squared
ggplot(daily_activity, aes(x = TotalDistance, y = Calories)) +
geom_point(alpha = 0.6, color = "blue") +
geom_smooth(method = "lm", color = "red", linewidth = 1) +
labs(title = "A Positive Correlation Between Daily Distance and Calories",
x = "Total Distance",
y = "Calories") +
annotate("text", x = max(daily_activity$TotalDistance) * 0.7,
y = max(daily_activity$Calories) * 0.9,
label = paste("R-squared = ", round(distance_calories_r_squared, 3)),
color = "black", size = 5, fontface = "bold") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
```
## `geom_smooth()` using formula = 'y ~ x'
``` r
steps_calories_model <- lm(Calories ~ TotalSteps, data = daily_activity)
steps_calories_r_squared <- summary(steps_calories_model)$r.squared
ggplot(daily_activity, aes(x = TotalSteps, y = Calories)) +
geom_point(alpha = 0.6, color = "blue") +
geom_smooth(method = "lm", color = "red", linewidth = 1) +
labs(title = "Positive Correlation of Steps vs Calories",
x = "Total Steps",
y = "Calories") +
annotate("text", x = max(daily_activity$TotalSteps) * 0.7,
y = max(daily_activity$Calories) * 0.9,
label = paste("R-squared = ", round(steps_calories_r_squared, 3)),
color = "black", size = 5, fontface = "bold") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
```
## `geom_smooth()` using formula = 'y ~ x'
- Both these metrics have similar predictive correlations for calories
burned.
``` r
steps_distance_model <- lm(TotalDistance ~ TotalSteps, data = daily_activity)
steps_distance_r_squared <- summary(steps_distance_model)$r.squared
ggplot(daily_activity, aes(x = TotalSteps, y = TotalDistance)) +
geom_point(alpha = 0.6, color = "blue") +
geom_smooth(method = "lm", color = "red", linewidth = 1) +
labs(title = "Closely Correlated As Expected",
x = "Total Daily Steps",
y = "Total Daily Distance") +
annotate("text", x = max(daily_activity$TotalSteps) * 0.7,
y = max(daily_activity$TotalDistance) * 0.9,
label = paste("R-squared = ", round(steps_distance_r_squared, 3)),
color = "black", size = 5, fontface = "bold") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
```
## `geom_smooth()` using formula = 'y ~ x'
- Both daily steps and total distance are a good predictor for daily
calories burned.
``` r
average_hourly_steps_intensity_merged <- hourly_steps %>%
merge(hourly_intensity, by = "ActivityDateHour") %>%
group_by(hour = hour(ActivityDateHour)) %>%
summarise(
average_hourly_steps = mean(StepTotal, na.rm = TRUE),
average_hourly_intensity = mean(TotalIntensity, na.rm = TRUE)
) %>%
arrange(hour)
ggplot(average_hourly_steps_intensity_merged, aes(x = hour)) +
geom_line(aes(y = average_hourly_steps, color = "Average Hourly Steps"), linewidth = 1) +
geom_line(aes(y = average_hourly_intensity * 100, color = "Average Hourly Intensity"), linewidth = 1) +
scale_y_continuous(
name = "Average Hourly Steps",
sec.axis = sec_axis(~ . / 100, name = "Average Hourly Intensity (Scaled x100)")
) +
labs(
title = "People Are Most Active Mid-day",
x = "Hour of Day"
) +
theme_minimal() +
theme(
axis.text.x = element_text(angle = 45, hjust = 1),
axis.title.y.right = element_text(color = "blue"),
axis.text.y.right = element_text(color = "blue"),
axis.title.y.left = element_text(color = "green"),
axis.text.y.left = element_text(color = "green")
) +
scale_color_manual(values = c("Average Hourly Steps" = "green", "Average Hourly Intensity" = "blue")) +
theme(legend.title = element_blank())
```
- Both hourly steps and intensity follow a similar trend of ramping up
mid-day.
``` r
percent_time_in_bed_sleeping <- sleep_day %>%
summarise( sum_total_minutes_asleep = sum(TotalMinutesAsleep, na.rm = TRUE),
sum_total_time_in_bed = sum(TotalTimeInBed, na.rm = TRUE),
percentage_asleep = (sum_total_minutes_asleep / sum_total_time_in_bed) * 100,
percentage_awake = 100 - percentage_asleep)
percent_sleep_pie <- data.frame(
category = c("Time In Bed Asleep", "Time In Bed Awake"),
percentage = c(percent_time_in_bed_sleeping$percentage_asleep,
percent_time_in_bed_sleeping$percentage_awake) )
ggplot(percent_sleep_pie, aes(x = "", y = percentage, fill = category)) +
geom_bar(stat = "identity", width = 1) +
coord_polar(theta = "y") +
labs(title = "Are Users Wasting Time Lounging Around In Bed?") +
theme_void() + # Remove background grid
scale_fill_manual(values = c("skyblue", "salmon")) +
geom_text(aes(label = paste0(round(percentage, 1), "%")),
position = position_stack(vjust = 0.5),
color = "black", size = 6)
```
- Here we have percent time spent sleeping as a function of total time
spent in bed, on average between all individual users.
- As a whole, it looks like users are not getting distracted and
having a hard time falling asleep.
- If they did spend more time awake in bed, we could possibly
implement a feature to try and remind users to stop all activity and
focus on trying to sleep if they want to hit their scheduled hours
of sleep goal, for those who need or want it.
# **Act -**
## Recommendations & final thoughts:
Marketing and branding:
- Work on a marketing campaign that reinvents our brand as more than
just a product, but as a lifestyle. That if you use our product,
you’re a part of an exclusive group with a company that knows and
understands you. A brand that empowers a group whose concerns have
been historically overlooked. We offer women focused recommendations
that no other company such as Fitbit can.
Expand products offerings:
- From analyzing our Fitbit smart device usage frequency, we saw that
29/33 users were categorized as “high frequency” users in which they
used their device over ~2/3 the individual days of the total study
period. They are clearly interested in the smart device tech field,
so we can market to and offer them other similar products as well.
We would conduct market research to figure out what this demographic
would also be interested in and go from there.
- By offering more products, a strong selling point to purchase more
of our products would be seamless integration across products from
our brand ecosystem, as well as possible cost savings under one
subscription membership. This would incentivize customers to stay
loyal to our brand, rather than exploring our competitors.
Expand current product features:
- Research other features that customers in this segment would be
into, and determine whether to offer it as a new product or add into
an existing product.
- Offer a connectedness feature. Give users the opportunity to share
to social media, or other “Bellabeat friends” when they achieve a
certain goal or milestone.
- Add an “educational feature” that gives recommendations from
credible sources such as healthy lifestyle habits, such as the
Center for Disease Control (CDC) or National Institute of Health
(NIH). We saw from our analysis that a good portion of adults from
our study barely met the daily recommended amount of steps and sleep
hours on the lower end.
- For recommendations such as the above, implement a reminder feature
that allows users to set a goal, with appropriate reminders to help
users reach that goal if it appears they are not on track.
Diversity distribution channels:
- Bellabeat currently only sells their products through online
retailers and their own website. An easy way to increase our reach
and market share is to extend our distribution channels. We could
begin offering our products in physical retail stores as well.
Bellabeat being a women-centric company, could further capitalize on
this by partnering with other women focused retail stores to reach
their target demographic easier.
Future analysis:
- If given the opportunity, I’d like to run a future analysis on
Bellabeat’s data itself, instead of relying on Fitbit’s data and
extrapolating to our company, as the differing demographics may be a
point of weak validity of this analysis. By analyzing Bellabeat’s
data, we would also get insight into some metrics that only our
women-centric products captured, and not the Fitbit, such as the
reproductive cycle.
- Trends for various metrics may also be different for women vs a
mixed, unknown demographic such as that of the Fitbit data.