Set options

knitr::opts_chunk$set(echo = TRUE, message = FALSE, warning = FALSE,
                      comment = NA)

Load required packages

library(dplyr)
library(lubridate)
library(ggplot2)
library(knitr)
library(plotly)

Load the data

activity <- read.csv("activity.csv",header = TRUE)
## transform in date
activity$date <- ymd(activity$date)

What is mean total number of steps taken per day?

Histogram of the total number of steps taken each day

bydate = activity %>%
    group_by(date) %>%
    summarise(N = n(), Total_steps = sum(steps, na.rm = TRUE))

ggplot(bydate, aes(x = Total_steps)) + geom_histogram(fill = "#A6CEE3", bins = 30, col = "#1F78B4") + ggtitle("Total number of steps Taken each day")

Mean and median number of steps taken each day

steps_summary <- bydate %>%
    summarise(Mean_steps = mean(Total_steps, na.rm = TRUE),
              Median_steps = median(Total_steps, na.rm = TRUE))

kable(steps_summary, align = 'l')
Mean_steps Median_steps
9354.23 10395

What is the average daily activity pattern?

Time series plot of the average number of steps taken

by_interval = activity %>%
    group_by(interval) %>%
    summarise(Mean_steps = mean(steps, na.rm = TRUE))
    
ggplot(by_interval, aes( x = interval, y = Mean_steps)) + geom_path() + labs(x = "5-minute intervals", title = "Average of steps taken across 5-minutes interval")

These are the top 10 intervals for the average steps

knitr::kable(activity %>%
    group_by(interval) %>%
    summarise(Mean_steps = mean(steps, na.rm = TRUE)) %>%
    group_by(interval) %>%
    summarise(Average_steps = mean(Mean_steps)) %>%
    arrange(desc(Average_steps)) %>%
    head(10), align = 'l')
interval Average_steps
835 206.1698
840 195.9245
850 183.3962
845 179.5660
830 177.3019
820 171.1509
855 167.0189
815 157.5283
825 155.3962
900 143.4528

Imputing missing values

Code to describe and show a strategy for imputing missing data

## Total number of Rows with missing values
sum(apply(activity, 1, FUN = function(x) {sum(is.na(x))}))
[1] 2304
#### Replace missing values with the mean for the interval ####

##create a merged dataset
new_df <- merge(activity, by_interval, by = "interval")

## substitute values
new_df$steps[is.na(new_df$steps)] <-  new_df$Mean_steps[is.na(new_df$steps)] 

new_df <- new_df %>%
    arrange(date) %>%
    select(steps, date, interval)

new_df$steps <- round(new_df$steps, digits = 0)

## the two datasets are equal but the new_df has no missing data
kable(head(new_df, 10), align = 'l')
steps date interval
2 2012-10-01 0
0 2012-10-01 5
0 2012-10-01 10
0 2012-10-01 15
0 2012-10-01 20
2 2012-10-01 25
1 2012-10-01 30
1 2012-10-01 35
0 2012-10-01 40
1 2012-10-01 45 
kable(head(activity, 10), align = 'l')
steps date interval
NA 2012-10-01 0
NA 2012-10-01 5
NA 2012-10-01 10
NA 2012-10-01 15
NA 2012-10-01 20
NA 2012-10-01 25
NA 2012-10-01 30
NA 2012-10-01 35
NA 2012-10-01 40
NA 2012-10-01 45

Histogram of the total number of steps taken each day after missing values are imputed

by_date_new <- new_df %>%
    group_by(date) %>%
    summarise(N = n(), Total_steps = sum(steps, na.rm = TRUE))

ggplot(by_date_new, aes(x = Total_steps)) + geom_histogram(fill = "#A6CEE3", bins = 30, col = "#1F78B4") + ggtitle("Total number of steps Taken each day", subtitle = "No missing data")

steps_summary_new <- by_date_new %>%
    summarise(Mean_steps = mean(Total_steps, na.rm = TRUE),
              Median_steps = median(Total_steps, na.rm = TRUE))

kable(steps_summary_new, align = 'l')
Mean_steps Median_steps
10765.64 10762

Mean and Median are not the same. They have become higher, as expected.

Are there differences in activity patterns between weekdays and weekends?

new_df$Weekday <- weekdays(new_df$date)
new_df$Week_part <- ifelse(new_df$Weekday %in% c("Sabato", "Domenica"), "Weekend", "Weekday")
head(new_df)
  steps       date interval Weekday Week_part
1     2 2012-10-01        0  Lunedì   Weekday
2     0 2012-10-01        5  Lunedì   Weekday
3     0 2012-10-01       10  Lunedì   Weekday
4     0 2012-10-01       15  Lunedì   Weekday
5     0 2012-10-01       20  Lunedì   Weekday
6     2 2012-10-01       25  Lunedì   Weekday

Panel plot comparing the average number of steps taken per 5-minute interval across weekdays and weekends

n = new_df %>%
    group_by(interval, Week_part) %>%
    summarise(mean_score = mean(steps))

ggplot(n, aes(x = interval, y = mean_score)) + geom_path() +
    facet_wrap(~Week_part, nrow = 2) + ylab("Number of Steps") + theme_light()