PA1_template.md

Reproducible Research: Peer Assessment 1

Loading data

Load and transform the data.

library(ggplot2)
library(plyr)
data <- read.csv("activity.csv")
data$date <- as.Date(data$date,format="%Y-%m-%d")

Mean total number of steps taken per day

We can plot the histogram of all the steps

with(data,hist(steps, main="Histogram of total steps"))


But would be better to see total steps per day. For that, we need to aggregate the data.

First we need to aggregate number of steps per day

data.perday <- aggregate(steps ~ date, data, sum, na.rm=T)

Now we can see histogram of steps per day

ggplot(data.perday, aes(steps))+geom_histogram(binwidth=2000,fill="white",colour="black")

Mean and media steps per day

mean(data.perday$steps,na.rm=T)

## [1] 10766

median(data.perday$steps,na.rm=T)

## [1] 10765

Average daily activity pattern

Now we aggregate steps per interval

data.interval <- aggregate(steps ~ interval, data, mean, na.rm=T)
ggplot(data.interval, aes(x=data.interval$interval, y=data.interval$steps))+geom_line()

Imputing missing values

We can impute the missing values by mean values

data.impute <- adply(data, 1, function(x) if (is.na(x$steps)) {
    x$steps = round(data.interval[data.interval$interval == x$interval, 2])
    x
} else {
    x
})

Now lets look at the per day and per interval patterns again after imputing missing values

First we need to aggregate number of steps per day

data.impute.perday <- aggregate(steps ~ date, data.impute, sum)

Now we can see histogram of steps per day

ggplot(data.impute.perday, aes(steps))+geom_histogram(binwidth=2000,fill="white",colour="black")

Mean and media steps per day

mean(data.impute.perday$steps)

## [1] 10766

median(data.impute.perday$steps)

## [1] 10762

Now we aggregate steps per interval

data.impute.interval <- aggregate(steps ~ interval, data.impute, mean)
ggplot(data.impute.interval, aes(x=interval, y=steps))+geom_line()

Are there differences in activity patterns between weekdays and weekends?

First find whether each day is a weekday or weekend day

data.impute.weekend <- subset(data.impute, weekdays(date) %in% c("Saturday", "Sunday"))
data.impute.weekday <- subset(data.impute, !weekdays(date) %in% c("Saturday", "Sunday"))

data.impute.weekend <- aggregate(steps ~ interval, data.impute.weekend, mean)
data.impute.weekday <- aggregate(steps ~ interval, data.impute.weekday, mean)

data.impute.weekend <- cbind(data.impute.weekend, day = rep("Weekend"))
data.impute.weekday <- cbind(data.impute.weekday, day = rep("Weekday"))


data.impute.week <- rbind(data.impute.weekend, data.impute.weekday)

Now plot the data

ggplot(data.impute.week, aes(x = interval, y = steps)) + geom_line() + facet_grid(day ~ 
    .) + labs(x = "Interval", y = "Number of steps")