PA1_template.Rmd

# Reproducible Research: Peer Assessment 1

```{r setoptions, echo=FALSE}
opts_chunk$set(echo = TRUE)
```

## Loading and preprocessing the data

Load data file.

```{r}
D <- read.csv("activity.csv",
        colClasses = c("integer", "Date", "integer"))
str(D)
head(D)
```

## What is mean total number of steps taken per day?

A histogram of the total number of steps taken each day.

```{r meanstep}
D.step <- tapply(D$steps, D$date, sum, na.rm = TRUE)
hist(D.step,
    xlab = "Total number of steps taken each days",
    ylab = "Number of days",
    breaks = 10)
```

```{r MeanAndMedian}
D.sum <- summary(D.step)
str(D.sum)
```

The mean and median total number of steps taken
per day are respectively `r D.sum[["Mean"]]` and `r as.integer(D.sum[["Median"]])`.

## What is the average daily activity pattern?

Time series plot of the average steps taken in each 5-minute interval across all the days.

```{r dailyactivityplot}
D.interval <- tapply(D$steps, D$interval, mean, na.rm = TRUE)
str(D.interval)
plot(D.interval, type = "l",
    xlab = "5-minute interval", ylab = "average of steps taken")
```

Which 5-minute interval on avarage contains the maximum?

```{r maximuminterval}
D.asc <- D.interval[order(-D.interval)]
D.maxintvl <- D.asc[1]
```

The 5-minute interval of `r names(D.maxintvl)` contains `r D.maxintvl` as the maximum value.

## Imputing missing values

```{r totalna}
D.na <- sum(is.na(D$steps))
```

The total number of missing values is `r D.na`.

Assign median of 5-minute interval to missing values of data. 

```{r missingvaluereplacment}
D.medianintvl <- tapply(D$steps, D$interval, median, na.rm = TRUE)
D.filter <- D
for (i in names(D.medianintvl)) {
    D.filter$steps <- replace(
                    D.filter$steps,
                    which(is.na(D.filter$steps)),
                    D.medianintvl[[i]])
}
```

The histogram of the total number of steps taken each days.

```{r histosumoffiltered}
D.step2 <- tapply(D.filter$steps, D.filter$date, sum)
hist(D.step2,
    xlab = "Total number of steps taken each days",
    ylab = "Number of days",
    breaks = 10)
```

The mean and median total number of steps taken per days.
```{r differencefromfilter}
D.mean <- tapply(D$steps, D$date, mean)
D.median <- tapply(D$steps, D$date, median)
D.mean2 <- tapply(D.filter$steps, D.filter$date, mean)
D.median2 <- tapply(D.filter$steps, D.filter$date, median)
hist(D.mean, breaks = 10)
hist(D.mean2, breaks = 10)
```

When we compare two histogram of median total number of steps taken per days, we found that higher distribution around steps of 0 indicates that most of the missing values are treated as 0 for median value. Imputing missing values changes the distribution to higher density near 0.

## Are there differences in activity patterns between weekdays and weekends?

Add a new column for weekdays with two levels of "weekday" and "weekend".

```{r weekdays}
D.filter$week <- !(weekdays(D.filter$date, abbreviate = TRUE) %in% c("土","日"))
D.filter$week <- factor(D.filter$week,
                    levels = c(TRUE, FALSE),
                    labels = c("weekday", "weekend"))
head(D.filter)
```

The time series plot of 5-minute interval and steps taken all the weekdays and weekends.

```{r weekdayplot, fig.height=10}
library(lattice)
D.aggr<- aggregate(steps ~ interval + week, D.filter, mean)
head(D.aggr)
xyplot(steps ~ interval | week, data = D.aggr, type = "l",
    xlab = "5-minute interval", ylab = "average steps taken")
```