groupdata2 (version 2.0.5)

downsample: Downsampling of rows in a data frame

Description

lifecycle::badge("maturing")

Uses random downsampling to fix the group sizes to the smallest group in the data.frame.

Wraps balance().

Usage

downsample(data, cat_col, id_col = NULL, id_method = "n_ids")

Value

data.frame with some rows removed. Ordered by potential grouping variables, `cat_col` and (potentially) `id_col`.

Arguments

data

data.frame. Can be grouped, in which case the function is applied group-wise.

cat_col

Name of categorical variable to balance by. (Character)

id_col

Name of factor with IDs. (Character)

IDs are considered entities, e.g. allowing us to add or remove all rows for an ID. How this is used is up to the `id_method`.

E.g. If we have measured a participant multiple times and want make sure that we keep all these measurements. Then we would either remove/add all measurements for the participant or leave in all measurements for the participant.

N.B. When `data` is a grouped data.frame (see dplyr::group_by()), IDs that appear in multiple groupings are considered separate entities within those groupings.

id_method

Method for balancing the IDs. (Character)

"n_ids", "n_rows_c", "distributed", or "nested".

n_ids (default)

Balances on ID level only. It makes sure there are the same number of IDs for each category. This might lead to a different number of rows between categories.

n_rows_c

Attempts to level the number of rows per category, while only removing/adding entire IDs. This is done in 2 steps:

1. If a category needs to add all its rows one or more times, the data is repeated.

2. Iteratively, the ID with the number of rows closest to the lacking/excessive number of rows is added/removed. This happens until adding/removing the closest ID would lead to a size further from the target size than the current size. If multiple IDs are closest, one is randomly sampled.

distributed

Distributes the lacking/excess rows equally between the IDs. If the number to distribute can not be equally divided, some IDs will have 1 row more/less than the others.

nested

Calls balance() on each category with IDs as cat_col.

I.e. if size is "min", IDs will have the size of the smallest ID in their category.

Author

Ludvig Renbo Olsen,

Details

Without `id_col`

Downsampling is done without replacement, meaning that rows are not duplicated but only removed.

With `id_col`

See `id_method` description.

See Also

Other sampling functions: balance(), upsample()

Examples

# Attach packages
library(groupdata2)

# Create data frame
df <- data.frame(
  "participant" = factor(c(1, 1, 2, 3, 3, 3, 3, 4, 4, 5, 5, 5, 5)),
  "diagnosis" = factor(c(0, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0)),
  "trial" = c(1, 2, 1, 1, 2, 3, 4, 1, 2, 1, 2, 3, 4),
  "score" = sample(c(1:100), 13)
)

# Using downsample()
downsample(df, cat_col = "diagnosis")

# Using downsample() with id_method "n_ids"
# With column specifying added rows
downsample(df,
  cat_col = "diagnosis",
  id_col = "participant",
  id_method = "n_ids"
)

# Using downsample() with id_method "n_rows_c"
# With column specifying added rows
downsample(df,
  cat_col = "diagnosis",
  id_col = "participant",
  id_method = "n_rows_c"
)

# Using downsample() with id_method "distributed"
downsample(df,
  cat_col = "diagnosis",
  id_col = "participant",
  id_method = "distributed"
)

# Using downsample() with id_method "nested"
downsample(df,
  cat_col = "diagnosis",
  id_col = "participant",
  id_method = "nested"
)