R：从数据帧计算加速因子

Question

I have the following data frame stored:

Source: local data frame [18 x 3]
Groups: instance [?]

   instance          V2             wtime
     (fctr)      (fctr)             (dbl)
1    CCRG10  BranchDBMS         2.1845122
2    CCRG10  CacheDBMS          0.8619093
3    CCRG20  BranchDBMS         7.3522605
4    CCRG20  CacheDBMS          2.5523066
5    CCRG30  BranchDBMS        15.7318869
6    CCRG30  CacheDBMS          5.1411876
7    CCRG40  BranchDBMS        31.7315724
8    CCRG40  CacheDBMS          7.6714212
9    CCRG50  BranchDBMS        58.0909133
10   CCRG50  CacheDBMS         11.3979914
11   CCRG60  BranchDBMS        78.5095645
12   CCRG60  CacheDBMS         15.5988044
13   CCRG70  BranchDBMS        94.0637485
14   CCRG70  CacheDBMS         20.2977642
15   CCRG80  BranchDBMS       102.8716548
16   CCRG80  CacheDBMS         25.0142898
17   CCRG90  BranchDBMS       100.5247555
18   CCRG90  CacheDBMS         28.3753977

I want to transform this table into a new one, eg

Source: local data frame [9 x 2]
Groups: instance [?]

   instance           speedup
     (fctr)             (dbl)
1    CCRG10         2.5345035
...

That is for each instance, I want to divide wtime for BranchDBMS by CacheDBMS , here 2.18/0.86=2.53.

How do I automate this process?

Answer 1

By looking at the posted output it seems that you manage your table within dplyr so tidyr approach would be a natural choice.

Code

Vectorize(require)(package = c("dplyr", "magrittr", "tidyr"),
                   character.only = TRUE)
dta %<>%
    spread(key = V3, value = V4) %>% 
    mutate(wtimRes = BranchDBMS / CacheDBMS) %>% 
    rename(instance = V2)

Results

> head(dta, 5)
  instance BranchDBMS  CacheDBMS  wtimRes
1   CCRG10   2.184512  0.8619093 2.534504
2   CCRG20   7.352260  2.5523066 2.880634
3   CCRG30  15.731887  5.1411876 3.059971
4   CCRG40  31.731572  7.6714212 4.136336
5   CCRG50  58.090913 11.3979914 5.096592

Gather

Naturally, if desired you may wish to gather your results into one column.

dta %<>%
    gather(key = key, value = value, -instance)

which would produce:

> head(dta,6)
  instance        key     value
1   CCRG10 BranchDBMS  2.184512
2   CCRG20 BranchDBMS  7.352260
3   CCRG30 BranchDBMS 15.731887
4   CCRG40 BranchDBMS 31.731572
5   CCRG50 BranchDBMS 58.090913
6   CCRG60 BranchDBMS 78.509564

---

Data import

dtaTxt <- "   instance          V2             wtime
     (fctr)      (fctr)             (dbl)
1    CCRG10  BranchDBMS         2.1845122
2    CCRG10  CacheDBMS          0.8619093
3    CCRG20  BranchDBMS         7.3522605
4    CCRG20  CacheDBMS          2.5523066
5    CCRG30  BranchDBMS        15.7318869
6    CCRG30  CacheDBMS          5.1411876
7    CCRG40  BranchDBMS        31.7315724
8    CCRG40  CacheDBMS          7.6714212
9    CCRG50  BranchDBMS        58.0909133
10   CCRG50  CacheDBMS         11.3979914
11   CCRG60  BranchDBMS        78.5095645
12   CCRG60  CacheDBMS         15.5988044
13   CCRG70  BranchDBMS        94.0637485
14   CCRG70  CacheDBMS         20.2977642
15   CCRG80  BranchDBMS       102.8716548
16   CCRG80  CacheDBMS         25.0142898
17   CCRG90  BranchDBMS       100.5247555
18   CCRG90  CacheDBMS         28.3753977"

dta <- read.table(textConnection(dtaTxt), header = FALSE, 
                  colClasses=c("NULL", NA, NA, NA), skip = 2)