PA1_template.md

Reproducible Research: Peer Assessment 1

Loading and preprocessing the data

library(lubridate)
library(ggplot2)
library(dplyr)
library(timeDate)
options(scipen = 5, digits = 0) #set number display options for knitr.

#Unpack and load activity.zip
if(file.exists("activity.csv") == F) {unzip("activity.zip")}
indata <- read.csv("activity.csv", header =T, sep=",")
indata$date <- ymd(indata$date)

What is mean total number of steps taken per day?

totals <- indata %>% filter(complete.cases(.)) %>% group_by(date) %>% 
    mutate(interval = NULL) %>% summarise_each(funs(sum))
ggplot(data = totals) + geom_histogram(aes(x=steps))

steps_mean <- mean(totals$steps)
steps_median <- median(totals$steps)

Mean number of steps taken each day is 10766 and median is 10765.

What is the average daily activity pattern?

totals <- indata %>% filter(complete.cases(.)) %>% group_by(interval) %>% 
   summarise(IntervalAverage=mean(steps))
ggplot(data = totals) + geom_line(aes(x=interval, y=IntervalAverage)) +
    ylab("Average number of steps")

interval_max <- totals[which.max(totals$IntervalAverage),1]

Maximum number of steps taken on average across all days contains interval 835.

Imputing missing values

Calculate interval average across all days and substitute missing values.

na_number <- sum(is.na(indata$steps))
#Calculate average for each interval
int_avg <- indata %>% group_by(interval) %>% 
    summarise(IntervalAverage=mean(steps, na.rm=T))

#Replace missing values in dataset
na_replace <- indata %>% mutate(steps = ifelse(is.na(steps), int_avg$IntervalAverage[match(interval, int_avg$interval)], steps) )

#Make histogram
totals <- na_replace %>% group_by(date) %>% mutate(interval = NULL) %>% summarise_each(funs(sum))
ggplot(data = totals) + geom_histogram(aes(x=steps))

steps_mean_adj <- mean(totals$steps)
steps_median_adj <- median(totals$steps)

Dataset contains 2304 missing values and that is 13% of the whole dataset. After filling in missing data, mean number of steps taken each day is still 10766 and median is now 10766.

Are there differences in activity patterns between weekdays and weekends?

totals <- na_replace %>% mutate(weekend = isWeekend(date)) %>%
    group_by(weekend, interval) %>% summarise(IntervalAverage=mean(steps))
totals$weekend[totals$weekend==T] <- "Weekend"
totals$weekend[totals$weekend==F] <- "Weekday"
ggplot(data = totals) + geom_line(aes(x=interval, y=IntervalAverage)) +
    facet_grid(weekend ~ .) + ylab("Average number of steps")