Package 'barn'

Combine datasets to preprocess

Description

Combine multiple data frames based on their common columns. That's the first step for preprocessing with the barn package. When printing it shows the characteristics of the combined datasets.

Usage

barn(..., nominal_sufix = "_cat", numeric_sufix = "_num")

## S3 method for class 'barn'
print(x, form_width = 30, ...)

Arguments

...	Extra arguments.
nominal_sufix	An optional string for dealing with nominal variables. Defaults to "_cat".
numeric_sufix	An optional string for dealing with numeric variables. Defaults to "_num".
x	An object of class "barn".
form_width	An integer specifying the minimum column width (in characters). Default is 30.

Value

A barn object containing the combined data frame, row counts,

Examples

full <- data.frame(id = 1:3, p1 = c("A", "B", "C"), p2 = 10:12, y = 1:3)
holdout <- data.frame(id = 4:5, p1 = c("D", "E"), p2 = 1:2)
original <- data.frame(id = 1:2, p1 = c("F", "G"), p2 = 3:4, y = 4:5)
print(barn(full, holdout, original))

---

Split the combined dataset

Description

Splits the combined data frame from a barn object back into a named list containing the preprocessed predictors.

Usage

harvest(barn_obj)

Arguments

barn_obj

An object of class "barn", created by barn().

Value

A named list of data frames, one for each dataset originally passed to barn().

Examples

full <- data.frame(id = 1:3, p1 = c("A", "B", "C"), p2 = 10:12, y = 1:3)
holdout <- data.frame(id = 4:5, p1 = c("D", "E"), p2 = 1:2)
original <- data.frame(id = 1:2, p1 = c("F", "G"), p2 = 3:4, y = 4:5)
harvested <- barn(full, holdout) |> harvest()
names(harvested)
harvested[["full"]]

---

Encode categorical columns with counts

Description

Frequency encoding of nominal variables.

Usage

plant_count_encode(barn_obj, nominal_suffix = "_cat")

Arguments

barn_obj	A Barn object, created by barn().
nominal_suffix	The suffix applied to column names. Defaults to "_cat".

Value

The modified barn_obj with the transformed combined data frame.

---

Extract decimals in numeric features

Description

Creates new integer columns by extracting specific digits from numeric columns. This function emulates a feature engineering technique often used in machine learning.

Usage

plant_decimals_extract(barn_obj, numeric_sufix = "_num", from = 1, to = 10)

Arguments

barn_obj	A barn object, created by barn().
numeric_sufix	The suffix used to identify numeric columns to process. Defaults to "_num".
from	The starting digit position to extract (e.g., 1 for the first decimal place). Defaults to 1.
to	The ending digit position to extract (e.g., 9 for the ninth decimal place). Defaults to 9.

Value

The modified barn_obj with the transformed combined data frame.

Examples

df <- tibble::tibble(x_num = c(1.234, 5.678, NA))
b <- barn(df) |> plant_decimals_extract(from = 1, to = 3)
harvest(b)[[1]]

---

Round numeric features to specified precisions

Description

Creates new numeric columns by rounding existing numeric columns at specified decimal precisions. This is useful for feature engineering, where different rounding granularities may capture meaningful patterns.

Usage

plant_decimals_round(barn_obj, numeric_sufix = "_num", precisions = c(9, 8))

Arguments

barn_obj	A barn object, created by barn().
numeric_sufix	The suffix used to identify numeric columns to process. Defaults to "_num".
precisions	A numeric vector specifying the number of decimal places to round to (e.g., c(9, 8)).

Value

The modified barn_obj with the transformed combined data frame.

Examples

df <- tibble::tibble(x_num = c(1.23456789))
b <- barn(df) |> plant_decimals_round(precisions = c(2, 3))
harvest(b)[[1]]
harvest(b)[[1]]$x_r2_num
harvest(b)[[1]]$x_r3_num

---

Encode labels in a barn object

Description

Transform nominal columns from factors to integers.

Usage

plant_label_encode(barn_obj)

Arguments

barn_obj

An instance of class "barn".

Value

The modified barn_obj with the transformed combined data frame.

---

Create new nominal pairs

Description

A function to create new features based on combinations of categorical columns in a barn object.

Usage

plant_new_nominal_pairs(barn_obj, nominal_suffix = "_cat")

Arguments

barn_obj	An object inheriting from the "barn" class.
nominal_suffix	A character string that specifies the suffix for the newly created columns. Optional, default is "_cat".

Value

The modified barn_obj with the transformed combined data frame.

---

New factors from numerical columns

Description

A function to transform numeric and character columns in a barn object into new factor columns. It appends "_num" for numeric columns, "_cat" for character columns, and renames both to factors. Original columns are deleted from the combined data frame within the barn object.

Usage

plant_new_numeric_factors(
  barn_obj,
  numeric_suffx = "_num",
  nominal_suffix = "_cat"
)

Arguments

barn_obj	A barn object, created by barn().
numeric_suffx	The suffix for new numeric factor columns. Default is "_num".
nominal_suffix	The suffix for new nominal factor columns. Default is "_cat".

Value

The modified barn_obj with the transformed combined data frame.

---

Summarize Barn Object

Description

A function to group and summarize to add aggregations to a barn_obj using specified variables and expressions. WARNING: Risk of overfitting and bad generalization if not done when resampling.

Usage

plant_summarize(barn_obj, .by = NULL, ...)

Arguments

barn_obj	An object of class 'barn'.
.by	Varible(s) to group by. Currently unused; must be empty.
...	Expressions to compute summarizing values.

Value

A modified barn_obj with summarized data in the combined slot.

---