Loading and preprocessing the data

Reading the data into activity_data and checking the type of data present. Converting the date from character to data type to make processing easier

        library(dplyr)
## 
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union
        library(ggplot2)
        activity_data <- read.csv("activity.csv", header=T, sep=',', na.strings="?", 
                                   nrows=17568, check.names=F, stringsAsFactors=F, comment.char="", quote='\"')
        
        activity_data$steps <- as.numeric(activity_data$steps)
## Warning: NAs introduced by coercion
        activity_data$date <- as.Date(activity_data$date, format="%Y-%m-%d")

What is mean total number of steps taken per day?

The following code aggregates the number of steps taken according to the date and draws a histogram using base plotting system. Further it prints the mean and median of the steps taken per day. 

        activity_steps_day <- aggregate(steps ~ date, data = activity_data, FUN = sum, na.rm = TRUE)
        hist(activity_steps_day$steps, main="Steps Per Day", 
             xlab="Number of Steps per day", ylab="Frequency", col="blue")

        print(paste("Mean of Number of Steps taken per day",mean(activity_steps_day$steps,na.rm = TRUE)))
## [1] "Mean of Number of Steps taken per day 10766.1886792453"
        print(paste("Median of Number of Steps taken per day",median(activity_steps_day$steps,na.rm = TRUE)))
## [1] "Median of Number of Steps taken per day 10765"

What is the average daily activity pattern?

The following code aggregates the number of steps taken according to the interval and draws a x-y plot using base plotting system. Further it prints the 5-minute interval which has the highest average number of steps 

        activity_steps_interval <- aggregate(steps ~ interval, data = activity_data, FUN = mean, na.rm = TRUE)
        with(activity_steps_interval, {
                plot(activity_steps_interval$steps~activity_steps_interval$interval, type="l",
                     ylab="Average Steps", xlab="Interval")
        })

        max_interval <- which(activity_steps_interval$steps==max(activity_steps_interval$steps))
        print(paste("5-minute interval which contains the maximum number of steps(average)",activity_steps_interval$interval[max_interval],
                    " with ",max(activity_steps_interval$steps)," steps."))
## [1] "5-minute interval which contains the maximum number of steps(average) 835  with  206.169811320755  steps."

Imputing missing values

The following code calculates the mean of the non missing values for each interval. It divides the activity data set into 2 data frames - one with all missing values and one with all non-missing values. After dividing each interval as a factor, the values calculated using non-missing values is added to the data frame of missing values for the corresponding interval using levels. The values are rounded off and converted to integer. The 2 data frame are thus combined back again. Updated number of missing values, mean and median is printed. A histogram containing the Number of Steps per day is drawn using the base plotting system

        print(paste("Total Number of NA Values in the dataset = ",sum(is.na(activity_data$steps))))
## [1] "Total Number of NA Values in the dataset =  2304"
        activity_steps_interval <- tapply(activity_data$steps, activity_data$interval, mean, na.rm = TRUE)
        missing_data <- activity_data[is.na(activity_data$steps),]
        non_missing_data <- activity_data[!is.na(activity_data$steps),]
        missing_data$steps <- as.factor(missing_data$interval)
        levels(missing_data$steps) <- activity_steps_interval
        levels(missing_data$steps) <- round(as.numeric(levels(missing_data$steps)))
        missing_data$steps <- as.integer(as.vector(missing_data$steps))
        
        combined_data <- rbind(missing_data, non_missing_data)
        
        print(paste("Total Number of NA Values in the new dataset = ",sum(is.na(combined_data$steps))))
## [1] "Total Number of NA Values in the new dataset =  0"
        activity_steps_day <- aggregate(steps ~ date, data = combined_data, FUN = sum, na.rm = TRUE)
        hist(activity_steps_day$steps, main="Steps Per Day", 
             xlab="Number of Steps per day", ylab="Frequency", col="blue")

        print(paste("Mean of Number of Steps taken per day",mean(activity_steps_day$steps,na.rm = TRUE)))
## [1] "Mean of Number of Steps taken per day 10765.6393442623"
        print(paste("Median of Number of Steps taken per day",median(activity_steps_day$steps,na.rm = TRUE)))
## [1] "Median of Number of Steps taken per day 10762"

Are there differences in activity patterns between weekdays and weekend

The following code creates a factor variable for weekdays and weekends. The total number of steps are calculated for each interval separately for weekend and weekday. ggplot is used to create this plot for weekend/weekday using facet_grid 

        weekdays_list <- c('Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday')
        activity_data$dayType <- factor((weekdays(activity_data$date) %in% weekdays_list), 
                           levels=c(FALSE, TRUE), labels=c('weekend', 'weekday'))
        activity_steps_day <- aggregate(steps ~ interval + dayType, data = activity_data, FUN = mean, na.rm = TRUE)
        names(activity_steps_day) <- c("interval", "dayType", "mean_steps")
        plot <- ggplot(activity_steps_day, aes(interval, mean_steps))
        plot + geom_line(color = "red") + facet_grid(dayType~.) +facet_wrap(dayType~.,ncol=1,strip.position="top")+ labs(x = "Intervals", y = "Average Number of Steps", title = "Activity on Weekends vs Weekdays")