Alexandre Huynen
3/12/2017
We start by unzipping and loading the data:
if(!file.exists("activity.csv")){
tmp <- tempfile()
download.file("https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2Factivity.zip", tmp, method = "curl")
unzip(tmp)
unlink(tmp)
}
activity <- read.csv("activity.csv")
str(activity)## 'data.frame': 17568 obs. of 3 variables:
## $ steps : int NA NA NA NA NA NA NA NA NA NA ...
## $ date : Factor w/ 61 levels "2012-10-01","2012-10-02",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ interval: int 0 5 10 15 20 25 30 35 40 45 ...
The variable date being encodded as a Factor, we convert it to the proper Date format:
activity$date <- ymd(activity$date)To make a histogram of the total number of steps taken each day, it is necessary to first process the data and group it by day. In the following R code, this is done using the dplyr package. The mean and median total number of steps taken per day are also calculated before plotting these results.
daily_activity <- activity %>% group_by(date) %>%
summarise(steps = sum(steps, na.rm = TRUE))
mean_steps <- mean(daily_activity$steps)
median_steps <- median(daily_activity$steps)
g1 <- ggplot(data = daily_activity, aes(x = steps)) +
geom_histogram(binwidth = 1000, alpha = 0.8,
fill = "light blue") +
geom_vline(aes(xintercept = mean_steps, colour = "Mean daily steps")) +
geom_vline(aes(xintercept = median_steps, colour = "Median daily steps")) +
labs(x = "Dayly number of steps", y = "Count",
title = "Histogram of the total number of steps taken each day") +
theme(legend.position = "bottom")
print(g1)
Observe that there are 10 days during which no activity is recorded.
The mean and the median are
print(mean_steps)## [1] 9354.23
print(median_steps)## [1] 10395
In order to analyze the daily activity pattern, we compute first the average number of steps taken during each 5-minute interval (accross all day). The resulting time series is then plotted.
int_activity <- activity %>% group_by(interval) %>%
summarise(avg_steps = mean(steps, na.rm = TRUE))
g2 <- ggplot(data = int_activity, aes(x = interval, y = avg_steps)) +
geom_line(color = "light blue", size = 1) +
labs(title = "Time series of the 5-minute interval average number of steps",
x = "Interval identifier", y = "Average number of steps taken")
print(g2)
The maximum numer of steps and the corresponding 5-minute interval are
filter(int_activity, avg_steps == max(avg_steps))## # A tibble: 1 × 2
## interval avg_steps
## <int> <dbl>
## 1 835 206.1698
As already mentioned, a considerable number of days/intervals are associated with missing values which may introduce bias into some calculations or summaries of the data. To circumvent this situation, we will now devise a simple strategy for filling in all of the missing values in the dataset.
The total number of missing values is
sum(is.na(activity$steps))## [1] 2304
The strategy adopted in this work consists in replacing the missing values by the average number of steps of the corresponding interval.
activityC <- activity %>%
transform(steps = ifelse(is.na(activity$steps),
int_activity$avg_steps[match(int_activity$interval,activity$interval)],
activity$steps
))
summary(activityC$steps)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.00 0.00 0.00 37.38 27.00 806.00
Using the new dataset activityC, we now make a histogram of the total number of steps taken each day,
daily_activityC <- activityC %>% group_by(date) %>%
summarise(steps = sum(steps, na.rm = TRUE))
mean_steps_C <- mean(daily_activityC$steps)
median_steps_C <- median(daily_activityC$steps)
g3 <- ggplot(data = daily_activityC, aes(x = steps)) +
geom_histogram(binwidth = 1000, alpha = 0.8,
fill = "light blue") +
geom_vline(aes(xintercept = mean_steps_C, colour = "Mean daily steps")) +
geom_vline(aes(xintercept = median_steps_C, colour = "Median daily steps")) +
labs(x = "Dayly number of steps", y = "Count",
title = "Histogram of the total number of steps taken each day (Cleaned dataset)") +
theme(legend.position = "bottom")
print(g3)
As it can be observed on the previous Figure, the mean and median daily steps calculated with the cleaned dataset increase and, additionally, have the same values.
print(mean_steps_C)## [1] 10766.19
print(median_steps_C)## [1] 10766.19
Also, as a result of the adopted strategy, the number of days with an average daily steps of about 11000 drastically increases (close to a 50% increase).
Unsing the cleaned dataset, we create a new Factor variable with two levels -- weekday and weekend indicating whether a given date is a weekday or weekend day. A panel plot containing a time series plot of the 5-minute interval (x-axis) and the average number of steps taken, averaged across all weekday days or weekend days (y-axis).
int_activityC <- activityC %>% mutate(daytype = ifelse(
weekdays(date) %in% c("Saturday", "Sunday"),
"Weekend", "Weekday")) %>% group_by(interval, daytype) %>%
summarise(avg_steps = mean(steps))
g4 <- ggplot(data = int_activityC, aes(x = interval, y = avg_steps)) +
geom_line(color = "light blue", size = 1) +
facet_grid(daytype~.) +
labs(title = "Time series of the 5-minute interval average number of steps (Cleaned dataset)",
x = "Interval identifier", y = "Average number of steps taken")
print(g4)
Finally, here are the mean number of steps taken by day type (i.e., Weekend or Weekday):
aggregate(avg_steps ~ daytype, int_activityC, mean)## daytype avg_steps
## 1 Weekday 35.61058
## 2 Weekend 42.36640