Supporting Statistical Analysis for Research

2.8 Subsetting rows

The tidyverse provides several functions to select rows from a tibble.

• filter() selects rows using a Boolean condition.

• sample_n() and sample_frac() take a random sample of the rows.

• slice() selects row by numeric position.

2.8.1 Examples

1. Create test and training data frames using filter().

   set.seed(145705)
   
   cps <-
     cps %>%
     mutate(
       split = ifelse(runif(n()) > .75, "test", "train")
     )
   cps_train <-
     cps %>%
     filter(split == "train")
   cps_test <-
     cps %>%
     filter(split == "test")
   
   dim(cps)    

   [1] 15992    11

   dim(cps_train)    

   [1] 11902    11

   dim(cps_test)    

   [1] 4090   11

2. Create test and training data frames using slice().

   set.seed(145705)
   test_indx <- which(runif(nrow(cps)) > .75)
   train_ind <- setdiff(1:nrow(cps), test_indx)
   
   cps_train <-
     cps %>%
     slice(train_ind)
   cps_test <-
     cps %>%
     slice(test_indx)
   
   dim(cps)    

   [1] 15992    11

   dim(cps_train)    

   [1] 11902    11

   dim(cps_test)    

   [1] 4090   11