The R Book

Selecting Rows from the Dataframe at Random

In bootstrapping or cross-validation we might want to select certain rows from the dataframe at random. We use the sample function to do this: the default replace = FALSE performs shuffling (each row is selected once and only once), while the option replace = TRUE (sampling with replacement) allows for multiple copies of certain rows. Here we use replace = F to select a unique 8 of the 20 rows at random:

worms[sample(1:20,8),]

            Field.Name    Area   Slope   Vegetation   Soil.pH     Damp   Worm.density
7         Church.Field     3.5       3    Grassland       4.2    FALSE              3
17           Cheapside     2.2       8        Scrub       4.7     TRUE              4
19          Gravel.Pit     2.9       1    Grassland       3.5    FALSE              1
14   Observatory.Ridge     1.8       6    Grassland       3.8    FALSE              0
12        North.Gravel     3.3       1    Grassland       4.1    FALSE              1
9          The.Orchard     1.9       0      Orchard       5.7    FALSE              9
11         Garden.Wood     2.9      10        Scrub       5.2    FALSE              8
8              Ashurst     2.1       0       Arable       4.8    FALSE              4

Note that the row numbers are in random sequence (not sorted), so that if you want a sorted random sample you will need to order the dataframe after the randomization.