activity.Rmd

Coursera - Reproducible Data Analysis - Assignment 1
==================================================

## Load and preprocess data

```{r load}
unzip('activity.zip')
data <- read.csv('activity.csv', 
                 colClasses = c('numeric','character','numeric'))
head(data,3)
```

-------
## Analysis the steps taken per day

```{r daystep, fig.width=5,fig.height=5}
# caculate the sum of steps taken each day
daystep <- tapply(data$steps, data$date,sum, na.rm = T)

# result
mean <- mean(daystep)
median <- as.integer(median(daystep)) # result show as integer
hist(daystep, col = 'blue',
     xlab = 'total steps each day',ylab = 'number of days')
```

**mean** total number of steps taken per day: `r mean`

**median** total number of steps taken per day: `r median`


-----
## Analysis average daily activity pattern

```{r intervalstep, fig.width=5,fig.height=5}
# analysis the steps by intervals
intervalstep <- tapply(data$steps, data$interval, mean, na.rm = T)

# result
max <- max(intervalstep)
plot(names(intervalstep),intervalstep, type= 'l', col = 'blue' ,
     xlab = '5-minute interval',   
     ylab = 'average steps across all day',   
     main = 'average daily activity pattern')
```

**maximum** average number of steps among all intervals: `r max` steps

-----
## Imputing missing values

### There are a number of days/intervals where data are missing.

```{r missingvalue, result = 'hide'}
bad <- is.na(data$steps)
num <- sum(bad)
```
**number of missing values** is: `r num`

### solution: impute missing values with average steps at that specific interval

```{r impute}
# get the index and the specific interval of missing values
badidx <- which(bad)
badinterval <- data[bad,]$interval

# impute and fill in
gooddata <- data
imputestep <- numeric()
n = 1
for(idx in badidx){
    itv <- badinterval[n]
    impute <- intervalstep[[as.character(itv)]]
    gooddata[idx,]$steps <- impute
    n <- n + 1  
}
head(gooddata,3)
```

### re-analysis the steps taken per day
```{r re-analysis, fig.width=5,fig.height=5}
# analysis the steps by intervals
daystep2 <- tapply(gooddata$steps, gooddata$date,sum)

# result
mean2 <- as.integer(mean(daystep2)) # result show as integer
median2 <- as.integer(median(daystep2)) 
hist(daystep, col = 'blue',
     xlab = 'total steps each day',ylab = 'number of days',
     main = 'histgram of daystep with imputed data')
```

Results differ from the estimates on unimputed data.

**new mean** of total number of steps taken per day: `r mean2`

**new median** of total number of steps taken per day: `r median2`

  

  
## differences in activity patterns between weekdays and weekends

```{r weekdays,results='hide'}
# get weekdays in English
Sys.setlocale('LC_TIME','C') # set weekday names in English
gooddata$date <- as.Date(gooddata$date)
gooddata$weekdays <- weekdays(gooddata$date)
```
```{r weekends}
# if it's weekend
weekends <- character()
for(d in 1:nrow(gooddata)){
  day <- gooddata[d,]$weekdays
  if(day %in% c('Sunday','Satruday')){
      weekends[d] <- 'weekend'
  }else{
      weekends[d] <- 'weekday'
  }
}
gooddata$weekends <- as.factor(weekends)

# split into weekday and weekend data.frame
w <- split(gooddata,gooddata$weekends)
dfweekday <- w$weekday
dfweekend <- w$weekend

# caculate average steps for each interval
itvstep.weekday <- tapply(dfweekday$steps, dfweekday$interval, mean)
itvstep.weekend <- tapply(dfweekend$steps, dfweekend$interval, mean)

# plot
par(mfrow=c(2,1),mar = c(4,4,1,2),oma = c(2,2,3,1))
plot(names(itvstep.weekday),itvstep.weekday, type= 'l',
     xlab = '', ylab = 'daily steps',col = 'blue')
legend("topright",legend='weekday')
plot(names(itvstep.weekend),itvstep.weekend, type= 'l',
     xlab = 'interval', ylab = 'daily steps',col = 'blue')
legend("topright",legend='weekend')
title(main = 'average daily activity pattern at each interval',outer = T)
```