From version 0.8.0, dplyr offers a handy function called group_split():
# On sample data from @Aus_10
df %>%
group_split(g)
[[1]]
# A tibble: 25 x 3
ran_data1 ran_data2 g
<dbl> <dbl> <fct>
1 2.04 0.627 A
2 0.530 -0.703 A
3 -0.475 0.541 A
4 1.20 -0.565 A
5 -0.380 -0.126 A
6 1.25 -1.69 A
7 -0.153 -1.02 A
8 1.52 -0.520 A
9 0.905 -0.976 A
10 0.517 -0.535 A
# … with 15 more rows
[[2]]
# A tibble: 25 x 3
ran_data1 ran_data2 g
<dbl> <dbl> <fct>
1 1.61 0.858 B
2 1.05 -1.25 B
3 -0.440 -0.506 B
4 -1.17 1.81 B
5 1.47 -1.60 B
6 -0.682 -0.726 B
7 -2.21 0.282 B
8 -0.499 0.591 B
9 0.711 -1.21 B
10 0.705 0.960 B
# … with 15 more rows
To not include the grouping column:
df %>%
group_split(g, keep = FALSE)