dplyr
Inhaltsverzeichnis
Introduction
The dplyr package makes each of these steps as fast and easy as possible by:
- Elucidating the most common data manipulation operations, so that your options are helpfully constrained when thinking about how to tackle a problem.
- Providing simple functions that correspond to the most common data manipulation verbs, so that you can easily translate your thoughts into code.
- Using efficient data storage backends, so that you spend as little time waiting for the computer as possible.
filter()
- Base R approach to filtering forces you to repeat the data frame’s name
- dplyr approach is simpler to write and read
- Command structure (for all dplyr verbs):
- first argument is a data frame
- return value is a data frame
- nothing is modified in place
- Note: dplyr generally does not preserve row names
filter() example
1 > require(dplyr)
2 > sub1 <- filter(data, Subject == 1)
3 > table(sub1$Subject)
4 1
5 665
6 > sub1 <- filter(data, Subject == 1, Stim.Type == "incorrect")
7 > table(sub1$Subject,sub1$Stim.Type)
8 hit incorrect other miss
9 1 0 293 0 0
10 > subframe <- filter(data, Age_PRETEST < 3.5 | Sex == "m" )
11 > table(subframe$Age_PRETEST < 3.5, subframe$Sex)
12 f m
13 FALSE 0 3202
14 TRUE 1333 1844
select()
- select() selects colums
- it can be used in combination with filter()
combining can be done by the infix %>%
select() example
1 > subframe <- select(data, Subject, Sex, Age_PRETEST)
2 > head(subframe)
3 Subject Sex Age_PRETEST
4 1 1 f 3.11
5 2 1 f 3.11
6 3 1 f 3.11
7 4 1 f 3.11
8 5 1 f 3.11
9 6 1 f 3.11
10 > subframe <- select(data, Subject, Sex, Age_PRETEST) %>%
11 + filter(Age_PRETEST < 3.2)
12 > table(subframe$Subject)
13 1 4 9 16 18
14 665 645 536 663 668
arrange()
- arrange can be used to change the order of rows
1 > arr.frame <- arrange(data, TTime, Time)
2 > head(arr.frame)
3 Subject Sex Age_PRETEST Trial Event.Type Code Time TTime Uncertainty
4 1 2 m 4.50 255 Response 1 9250486 2 1
5 2 15 f 4.11 381 Response 2 7850406 10 1
6 3 14 m 4.60 297 Response 1 11254989 13 1
7 4 17 m 4.90 234 Response 2 12267915 13 1
8 5 9 m 3.11 127 Response 1 1445239 16 1
9 6 2 m 4.50 332 Response 2 3580014 24 1
10 Duration Uncertainty.1 ReqTime ReqDur Stim.Type Pair.Index Type Event.Code
11 1 200 2 0 next hit 220 Picture TO18.jpg
12 2 200 2 0 next hit 328 Picture TO22.jpg
13 3 200 2 0 next hit 258 Picture TS05.jpg
14 4 200 2 0 next incorrect 202 Picture TO03.jpg
15 5 200 2 0 next hit 126 Picture RS21.jpg
16 6 200 2 0 next hit 333 Picture RS30.jpg
17 Time1 testid EC1
18 1 9250484 1 TO
19 2 7850396 4 TO
20 3 11254976 5 TS
21 4 12267902 6 TO
22 5 1445223 test2 RS
23 6 3579990 test2 RS
mutate()
- mutate() can be used to transform or add columns
- you can add/change more than one column at once
mutate() example
1 > subframe <- filter(data, Subject == 1) %>%
2 + mutate(Event.Code = str_replace(Event.Code,".jpg",""),
3 + TTime.calc = Time - Time1)
4 > head(subframe)
5 Subject Sex Age_PRETEST Trial Event.Type Code Time TTime Uncertainty
6 1 1 f 3.11 7 Response 2 103745 2575 1
7 2 1 f 3.11 12 Response 2 156493 2737 1
8 3 1 f 3.11 17 Response 2 214772 6630 1
9 4 1 f 3.11 22 Response 1 262086 5957 1
10 5 1 f 3.11 27 Response 2 302589 272 1
11 6 1 f 3.11 32 Response 1 352703 7197 1
12 Duration Uncertainty.1 ReqTime ReqDur Stim.Type Pair.Index Type Event.Code
13 1 2599 3 0 next hit 7 Picture RO26
14 2 2800 2 0 next incorrect 12 Picture RO19
15 3 6798 2 0 next hit 17 Picture RS23
16 4 5999 2 0 next incorrect 22 Picture OF22
17 5 400 2 0 next hit 27 Picture AT08
18 6 7398 2 0 next hit 32 Picture AT30
19 Time1 testid EC1 TTime.calc
20 1 101170 test2 RO 2575
21 2 153756 test2 RO 2737
22 3 208142 test2 RS 6630
23 4 256129 test2 OF 5957
24 5 302317 test2 AT 272
25 6 345506 test2 AT 7197
26 > table(subframe$Subject)
27 1
28 665
transmute()
- transmute() does the same like mutate{} but keeps only the new columns
summarise()
- summarise() makes summary statistics
summarise() example
- summarise() gets really interesting in combination with group\by() also included in the dplyr package
1 > sum.frame <- group_by(data, Subject) %>%
2 + summarise(mean.ttime=mean(TTime), sd.ttime = sd(TTime))
3 > sum.frame
4 Subject mean.ttime sd.ttime
5 1 1 11717.854 13035.85
6 2 2 13100.568 13607.71
7 3 3 15709.598 16464.09
8 4 4 24778.592 20205.91
9 5 5 14759.785 14863.84
10 6 6 14081.377 14834.64
11 7 7 11551.482 12814.57
12 8 8 22739.310 18215.68
13 9 9 20490.722 19399.49
summarise() example
1 > sum.frame <- group_by(data, Subject, testid) %>%
2 + summarise(mean.ttime=mean(TTime), sd.ttime = sd(TTime))
3 > head(sum.frame)
4 Subject testid mean.ttime sd.ttime
5 1 1 test1 8621.674 7571.462
6 2 1 1 9256.367 8682.833
7 3 1 2 9704.712 10479.788
8 4 1 3 14189.550 13707.021
9 5 1 4 13049.831 11344.656
10 6 1 5 14673.525 15575.355
dplyr Exercises
* use select() and filter() in combination ($>$) to select all rows belonging to the post or the pre test, keep the Subject, Sex, Age_PRETEST and Stim.Type column. Create a new data frame named data2 or something like this. * add two new variables containing the counts of hit and incorrect. Use mutate() and sum(Stim.Type=='hit'). * use group_by() and summarise() to extract the minimum and maximum TTime per person from the original data frame * repeat the last exercise, but now group per person and EC1
dplyr Exercise 1 Solution
use select() and filter() in combination (%>%) to select all rows belonging to the post or the pre test, keep the Subject, Sex, Age_PRETEST and Stim.Type column. Create a new data frame named data2 or something like this.
dplyr Exercises 2 Solution
- add two new variables containing the counts of hit and incorrect. Use mutate() and sum(Stim.Type=='hit')
1 > data2 <- mutate(data2,n.hit=sum(Stim.Type=='hit'),
2 + n.incorrect=sum(Stim.Type=='incorrect'))
3 > head(data2)
4 Subject Sex Age_PRETEST Stim.Type n.hit n.incorrect
5 1 1 f 3.11 hit 2561 1223
6 2 1 f 3.11 incorrect 2561 1223
7 3 1 f 3.11 hit 2561 1223
8 4 1 f 3.11 incorrect 2561 1223
9 5 1 f 3.11 hit 2561 1223
10 6 1 f 3.11 hit 2561 1223
dplyr Exercises 3 Solution
- use group_by() and summarise() to extract the minimum and maximum TTime per person from the original data frame
1 > sum.frame <- group_by(data, Subject) %>%
2 + mutate(min.ttime = min(TTime), max.ttime=max(TTime))
3 > head(sum.frame[,c(1:3,20:22)])
4 Subject Sex Age_PRETEST EC1 min.ttime max.ttime
5 1 1 f 3.11 RO 46 96434
6 2 1 f 3.11 RO 46 96434
7 3 1 f 3.11 RS 46 96434
8 4 1 f 3.11 OF 46 96434
9 5 1 f 3.11 AT 46 96434
10 6 1 f 3.11 AT 46 96434
dplyr Exercises 4 Solution
- repeat the last exercise, but now group per person and EC1
1 > sum.frame <- group_by(data, Subject, EC1) %>%
2 + mutate(min.ttime = min(TTime), max.ttime=max(TTime))
3 > head(sum.frame[,c(1:3,20:22)])
4 Subject Sex Age_PRETEST EC1 min.ttime max.ttime
5 1 1 f 3.11 RO 365 30510
6 2 1 f 3.11 RO 365 30510
7 3 1 f 3.11 RS 423 54085
8 4 1 f 3.11 OF 298 58939
9 5 1 f 3.11 AT 272 17344
10 6 1 f 3.11 AT 272 17344