Activity Data Analysis
Jake Ackman
10/12/2019
fileUrl <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2Factivity.zip"
download.file(fileUrl, destfile = paste0(getwd(), '/repdata%2Fdata%2Factivity.zip'))
unzip("repdata%2Fdata%2Factivity.zip",exdir = "data")
act <- read.csv("data/activity.csv")Activity Data
Read in the data above to act dataframe. Then below we:
-Calculate the average steps per day
-Chart a histogram of steps per day
-Calculate the mean and median steps
-Chart the mean steps per time interval
-Identify that 835 is the five minute time interval with the highest average
steps_per_day <- act %>% group_by(date) %>% dplyr::summarise(steps_per_day = sum(steps, na.rm = TRUE))
steps_per_day## # A tibble: 61 x 2
## date steps_per_day
## <fct> <int>
## 1 2012-10-01 0
## 2 2012-10-02 126
## 3 2012-10-03 11352
## 4 2012-10-04 12116
## 5 2012-10-05 13294
## 6 2012-10-06 15420
## 7 2012-10-07 11015
## 8 2012-10-08 0
## 9 2012-10-09 12811
## 10 2012-10-10 9900
## # ... with 51 more rowsggplot(steps_per_day, aes(steps_per_day)) + geom_histogram() + ggtitle("Steps Per Day Histogram")## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
act %>% filter(!is.na(steps)) %>% dplyr::group_by(date) %>% dplyr::summarise(mean_steps = mean(steps, na.rm = TRUE), median_steps = median(steps, na.rm =TRUE))## # A tibble: 53 x 3
## date mean_steps median_steps
## <fct> <dbl> <dbl>
## 1 2012-10-02 0.438 0
## 2 2012-10-03 39.4 0
## 3 2012-10-04 42.1 0
## 4 2012-10-05 46.2 0
## 5 2012-10-06 53.5 0
## 6 2012-10-07 38.2 0
## 7 2012-10-09 44.5 0
## 8 2012-10-10 34.4 0
## 9 2012-10-11 35.8 0
## 10 2012-10-12 60.4 0
## # ... with 43 more rowsact %>% filter(!is.na(steps)) %>% dplyr::group_by(interval) %>% dplyr::summarise(avg = mean(steps, na.rm = TRUE)) %>% ggplot(.,aes(interval, avg)) + geom_line() + ggtitle("Mean steps per Time Interval")
Impute NAs
Get rid of those dang NAs! First show the total number of NAs. Then create a new variable so that the mean replaces any NAs. Add a histogram of the imputed mean, and then show a table of the
sum(is.na(act$steps))## [1] 2304act_imputed <- act %>% mutate(impute = ifelse(is.na(steps), mean(steps, na.rm = TRUE), steps))
act_imputed %>% dplyr::group_by(date) %>% dplyr::summarise(mean_imputed = mean(impute)) %>% ggplot(.,aes(mean_imputed)) + geom_histogram() + ggtitle("Mean Steps Per Day From Dataset with Imputed Mean")## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
act_imputed %>% dplyr::group_by(date) %>% dplyr::summarise(avg = mean(impute, na.rm = TRUE), median = median(impute, na.rm = TRUE)) %>% kable()| date | avg | median |
|---|---|---|
| 2012-10-01 | 37.3825996 | 37.3826 |
| 2012-10-02 | 0.4375000 | 0.0000 |
| 2012-10-03 | 39.4166667 | 0.0000 |
| 2012-10-04 | 42.0694444 | 0.0000 |
| 2012-10-05 | 46.1597222 | 0.0000 |
| 2012-10-06 | 53.5416667 | 0.0000 |
| 2012-10-07 | 38.2465278 | 0.0000 |
| 2012-10-08 | 37.3825996 | 37.3826 |
| 2012-10-09 | 44.4826389 | 0.0000 |
| 2012-10-10 | 34.3750000 | 0.0000 |
| 2012-10-11 | 35.7777778 | 0.0000 |
| 2012-10-12 | 60.3541667 | 0.0000 |
| 2012-10-13 | 43.1458333 | 0.0000 |
| 2012-10-14 | 52.4236111 | 0.0000 |
| 2012-10-15 | 35.2048611 | 0.0000 |
| 2012-10-16 | 52.3750000 | 0.0000 |
| 2012-10-17 | 46.7083333 | 0.0000 |
| 2012-10-18 | 34.9166667 | 0.0000 |
| 2012-10-19 | 41.0729167 | 0.0000 |
| 2012-10-20 | 36.0937500 | 0.0000 |
| 2012-10-21 | 30.6284722 | 0.0000 |
| 2012-10-22 | 46.7361111 | 0.0000 |
| 2012-10-23 | 30.9652778 | 0.0000 |
| 2012-10-24 | 29.0104167 | 0.0000 |
| 2012-10-25 | 8.6527778 | 0.0000 |
| 2012-10-26 | 23.5347222 | 0.0000 |
| 2012-10-27 | 35.1354167 | 0.0000 |
| 2012-10-28 | 39.7847222 | 0.0000 |
| 2012-10-29 | 17.4236111 | 0.0000 |
| 2012-10-30 | 34.0937500 | 0.0000 |
| 2012-10-31 | 53.5208333 | 0.0000 |
| 2012-11-01 | 37.3825996 | 37.3826 |
| 2012-11-02 | 36.8055556 | 0.0000 |
| 2012-11-03 | 36.7048611 | 0.0000 |
| 2012-11-04 | 37.3825996 | 37.3826 |
| 2012-11-05 | 36.2465278 | 0.0000 |
| 2012-11-06 | 28.9375000 | 0.0000 |
| 2012-11-07 | 44.7326389 | 0.0000 |
| 2012-11-08 | 11.1770833 | 0.0000 |
| 2012-11-09 | 37.3825996 | 37.3826 |
| 2012-11-10 | 37.3825996 | 37.3826 |
| 2012-11-11 | 43.7777778 | 0.0000 |
| 2012-11-12 | 37.3784722 | 0.0000 |
| 2012-11-13 | 25.4722222 | 0.0000 |
| 2012-11-14 | 37.3825996 | 37.3826 |
| 2012-11-15 | 0.1423611 | 0.0000 |
| 2012-11-16 | 18.8923611 | 0.0000 |
| 2012-11-17 | 49.7881944 | 0.0000 |
| 2012-11-18 | 52.4652778 | 0.0000 |
| 2012-11-19 | 30.6979167 | 0.0000 |
| 2012-11-20 | 15.5277778 | 0.0000 |
| 2012-11-21 | 44.3993056 | 0.0000 |
| 2012-11-22 | 70.9270833 | 0.0000 |
| 2012-11-23 | 73.5902778 | 0.0000 |
| 2012-11-24 | 50.2708333 | 0.0000 |
| 2012-11-25 | 41.0902778 | 0.0000 |
| 2012-11-26 | 38.7569444 | 0.0000 |
| 2012-11-27 | 47.3819444 | 0.0000 |
| 2012-11-28 | 35.3576389 | 0.0000 |
| 2012-11-29 | 24.4687500 | 0.0000 |
| 2012-11-30 | 37.3825996 | 37.3826 |
Weekday Analysis
Created a factor variable called weekday to show if it’s a weekday or the weekend. Graphed the imputed mean per time interval for Weekdays vs. Weekends.
act_weekday <- act_imputed %>% dplyr::mutate(day = weekdays(ymd(date)))
act_weekday <- act_weekday %>% mutate(weekday = ifelse(day %in% c("Saturday", "Sunday"), "Weekend", "Weekday"))
act_weekday$weekday <- factor(act_weekday$weekday, levels = c("Weekday", "Weekend"))
ggplot(act_weekday, aes(interval, impute)) + geom_line(color = "blue") + facet_grid(weekday~., scales = "free") + theme_classic()