Which days have the least/most amount of steps?

#plotting Total Number of Steps according to Weekdays
weekday <- c("Monday", "Tuesday", "Wednesday", "Thursday", 
             "Friday", "Saturday", "Sunday")
ggplot(data=daily_activity, aes(x=weekday, y=total_steps)) + 
  geom_bar(stat="identity", fill='pink') + 
  labs(title='Total Number of Steps per Weekday') + 
  scale_x_discrete(limits = weekday) + 
  xlab("Weekdays") + ylab("Number of Steps")

By plotting the total number of steps per weekday, the most and the least active days recorded can be found. According to the above figure, Tuesday seems to be the most active day for all the users since it has the most steps. The least amount of total steps being on Sunday is not very surprising, since it is a rest day for most people. Following Sunday are Monday and Friday.
Based on these findings, users can be given extra motivation through the app to start the week on a more positive note, with increased activity. Same goes for Fridays, end of the week can be tiring however users can be encouraged to not give up, and keep moving by adding certain game-like characteristics such as achievements or a notification showing them which percentage of users they fall under based on their activity levels for the day.

#Calculating the average steps per weekday
average_weekday_activity <- daily_activity %>%
  group_by(weekday) %>% 
  summarise(average_steps_per_weekday = mean(total_steps))

#plotting the average steps per weekday
ggplot(data=average_weekday_activity, 
       aes(x=weekday, y=average_steps_per_weekday)) + 
  geom_bar(stat="identity",fill="pink") + 
  scale_x_discrete(limits = weekday) + 
  xlab("Weekdays") + ylab("Average Steps") + 
  labs(title="Average Steps per Weekday by All Users")

On average, users moved the most on Saturdays, followed by Tuesday. Sunday is the day with the least steps on average.

Which days have the least/most amount of minutes slept?

#plotting Total Minutes Slept per Weekday
ggplot(data=daily_sleep, aes(x=weekday, y=total_minutes_asleep)) + 
  geom_bar(stat='identity',fill="pink") + 
  scale_x_discrete(limits = weekday) + 
  xlab("Weekdays") + ylab("Total Minutes Slept") + 
  labs(title="Total Minutes Slept per Weekday")

Above plot shows the total minutes slept by all the users based on the weekday. Interestingly, Wednesday seems to be the day with the most minutes slept. The least total minutes slept however are on Monday, this can very well be due to not being able to sleep because of the anxiety of starting a new week or avoiding sleep to make the precious Sunday night last longer. On these days with the least sleep, it can be good to remind the users of benefit of sleep, and how much better they will feel in the morning and in the future with a fixed sleep schedule.

#Calculating average minutes slept per Weekday
average_weekday_sleep <- daily_sleep %>%
  group_by(weekday) %>% 
  summarise(average_sleep_per_weekday = mean(total_minutes_asleep))

#plotting average minutes slept per weekday
ggplot(data=average_weekday_sleep, aes(x=weekday, y=average_sleep_per_weekday)) + 
  geom_bar(stat="identity",fill="pink") + 
  scale_x_discrete(limits = weekday) + 
  xlab("Weekdays") + ylab("Average Sleep") + 
  labs(title="Average Minutes Slept per Weekday by All Users")

According to this graph Sunday is the day with the most minutes slept on average, followed by Wednesday. All the other days do not have that big of a difference between them.

Categorizing Users

Instead of analyzing every single user individually, the users can be categorized into three different groups based on their average activity. According to this article, the classification of the users can be done like this:

• <5000 steps/day, ‘sedentary’
• 5000-7499 steps/day, ‘low active’
• 7500-9999 steps/day, ‘somewhat active’
• “>or=10000 steps/day, ‘active’”
• “>12500 steps/day, ‘highly active’”

user_type <- daily_activity %>% 
  group_by (id) %>% 
  summarise(average_steps= mean(total_steps)) %>% 
  mutate(user_type= case_when(
    average_steps < 5000 ~"sedentary",
    average_steps >= 5000 & average_steps <7499 ~"low active",
    average_steps >= 7499 & average_steps <9999 ~"somewhat active",
    average_steps >= 10000 & average_steps <12500 ~"active",
    average_steps > 12500 ~"highly active"))

daily_activity <- merge(x= daily_activity, y=user_type, by='id')

Which activity level is the least popular?

activity_level <- c("sedentary", "low active", "somewhat active", 
                    "active", "highly active")
ggplot(data = daily_activity, aes(x = user_type)) +
  geom_bar(color = "purple", fill = "pink") + 
  labs(title = "Activity Levels") + 
  xlab("User Type") + ylab("Number of Users") + 
  scale_x_discrete(limits = activity_level)

Above figure visualizes the categorization of the number of users per activity level. Most of the users seem to belong to the activity level “somewhat active”, which was defined as taking 7500-9999 steps per day which is not a small amount. However, “sedentary” and “low active” users seem to be higher in quantity compared to “active” and “highly active” users.

What hour of the daytime is the least/most active?

The column activity_hour needs to be separated into date and time for analyzing the hourly steps.


hourly_steps <- hourly_steps %>% 
  separate(activity_hour, into = c("date", "time"), sep= " ")

Below code helps with calculating the average steps by all the users per hour of the day.

hourly_steps_average <- hourly_steps %>% 
  group_by(time) %>% 
  summarise(average_steps = mean(step_total))

Plotting above findings results in the below graph.

ggplot(hourly_steps_average, aes(x=time, y=average_steps, fill=average_steps)) + 
  geom_col() + theme(axis.text.x = element_text(angle = 90)) + 
  xlab("Time") + ylab("Average Steps") 

According to this graph, the hour in which people move the most on average is 6 pm, followed by 7 pm. This can be easily explained by commuting or going for after work walks. The other hours during the day where people move a lot are between 12-2 pm, which is also the time most people have lunch breaks.