r - 如何使用长格式的 R 数据帧的子集进行操作？

Question

我有一个包含 3 个组和 3 天的数据框:

set.seed(10)
dat <- data.frame(group=rep(c("g1","g2","g3"),each=3), day=rep(c(0,2,4),3), value=runif(9))
#   group day    value
# 1    g1   0 0.507478
# 2    g1   2 0.306769
# 3    g1   4 0.426908
# 4    g2   0 0.693102
# 5    g2   2 0.085136
# 6    g2   4 0.225437
# 7    g3   0 0.274531
# 8    g3   2 0.272305
# 9    g3   4 0.615829

我想取 log2 并将每个值除以每组中的第 0 天值。我现在的做法是在中间步骤中计算每个日组:

day_0 <- dat[dat$day==0, "value"]
day_2 <- dat[dat$day==2, "value"]
day_4 <- dat[dat$day==4, "value"]
res <- cbind(0, log2(day_2/day_0), log2(day_4/day_0))
rownames(res) <- c("g1","g2","g3")
colnames(res) <- c("day_0","log_ratio_day_2_day_0","log_ratio_day_4_day_0")
#    day_0 log_ratio_day_2_day_0 log_ratio_day_4_day_0
# g1     0            -0.7261955             -0.249422
# g2     0            -3.0252272             -1.620346
# g3     0            -0.0117427              1.165564

正确的计算方式是什么 res没有中间步骤？

Answer 1

一个 data.table编码优雅和内存效率的解决方案

library(data.table)

DT <- data.table(dat)

# assign within DT by reference

DT[, new_value := log2(value / value[day == 0]), by = group]

或者你可以使用 joins和 keys和 by-without-by

DTb <- data.table(dat)

setkey(DTb, group)

# val0 contains just those records for day 0
val0 <- DTb[day==0]

 # the i.value refers to value from the i argument 
 # which is in this case `val0` and thus the value for 
 # day = 0 
 DTb[val0, value := log2(value / i.value)]

这两种解决方案都不需要您按 day 排序确保 value将第一个(或任何特定的)元素。

编辑
文档为 i.句法

    **********************************************
    **                                          **
    **   CHANGES IN DATA.TABLE VERSION 1.7.10   **
    **                                          **
    **********************************************
     NEW FEATURES

o   New function setcolorder() reorders the columns by name
    or by number, by reference with no copy. This is (almost)
    infinitely faster than DT[,neworder,with=FALSE].
    
o   The prefix i. can now be used in j to refer to join inherited
    columns of i that are otherwise masked by columns in x with
    the same name.