grouping

# vectorise
group_f1 = function(data){
    new_col = cut(X[,8], c(-Inf, 0.25, 0.5, 0.75, Inf))
    levels(new_col) <- paste0("group", 1:4)
    cbind(data, new_col)
}

# for loop
group_f2 = function(data){
    new_col = rep(NA, nrow(data))
    for (i in 1:length(new_col)){
      if (data[i,8] < .25){
        new_col[i] = "group1"
      } else if (data[i,8] < .5){
        new_col[i] = "group2"
      } else if (data[i,8] < .75){
        new_col[i] = "group3"
      } else {
        new_col[i] = "group4"
      }
    }
    cbind(data, new_col)
}

# sapply
group_f3 = function(data){
    new_col = sapply(1:nrow(data), function(i){
      if (data[i,8] < .25){
        "group1"
      } else if (data[i,8] < .5){
        "group2"
      } else if (data[i,8] < .75){
        "group3"
      } else {
        "group4"
      }
    })
    cbind(data, new_col)
}

# plyr and dplyr
library(data.table)
library(plyr)
library(dplyr)
library(magrittr)
group_f4 = function(data){
  data %>% mutate(new_col = cut(X8, c(-Inf, 0.25, 0.5, 0.75, Inf))) %>%
    transform(new_col = mapvalues(new_col, from = levels(new_col),
      to = paste0("group", 1:4)))
}

library(rbenchmark)
## data generation
X = sapply((1:10-5.5)*4/20, rnorm, n = 200000)
X_df = data.frame(X) %>% tbl_df()
X_dt = data.table(X_df) %>% tbl_dt(FALSE)
benchmark(
  group_f1(X_df), group_f2(X_df), group_f3(X_df),
  group_f4(X_df), group_f4(X_dt),
  columns = c("test", "replications", "elapsed", "relative", "user.self"),
  order = "relative", replications = 20
)
#             test replications elapsed relative user.self
# 1 group_f1(X_df)           20    0.96    1.000      0.95
# 4 group_f4(X_df)           20    0.98    1.021      0.97
# 5 group_f4(X_dt)           20    1.23    1.281      1.20
# 2 group_f2(X_df)           20  417.14  434.521    411.56
# 3 group_f3(X_df)           20  418.21  435.635    412.50