fctutils: Advanced Factor Manipulation Utilities

Kai Guo, University of Michigan

2024-09-30

The fctutils package provides a comprehensive suite of utilities for advanced manipulation and analysis of factor vectors in R. It offers tools for splitting, combining, reordering, filtering, and transforming factor levels based on various criteria. Designed to enhance the handling of categorical data, fctutils simplifies complex factor operations, making it easier to preprocess and analyze data in R.

Key Features:

1. Software Usage

1.1 Installation

Install the package with its dependencies and load it for usage in R.

library(devtools) # Load the devtools package
install_github("guokai8/fctutils") # Install the package

2. Useful functions

2.1 Ordering and Sorting Factors

ft_pos Reorders the levels of a factor vector based on the characters at specified positions within the factor levels.

library(fctutils)
factor_vec <- factor(c('Apple', 'banana', 'Cherry', 'date', 'Fig', 'grape'))
# Reorder based on positions 1 and 3, case-insensitive
ft_pos(factor_vec, positions = c(1, 3))
## [1] Apple  banana Cherry date   Fig    grape 
## Levels: Apple banana Cherry date Fig grape
# Reorder based on positions 3, case-insensitive, inplace = TRUE
ft_pos(factor_vec, positions = 3, inplace = TRUE)
## [1] grape  Cherry Fig    banana Apple  date  
## Levels: grape Cherry Fig banana Apple date
# Reorder in decreasing order, case-sensitive
ft_pos(factor_vec, positions = 1:2, case = TRUE, decreasing = TRUE)
## [1] Apple  banana Cherry date   Fig    grape 
## Levels: grape date banana Fig Cherry Apple

ft_count Reorders the levels of a factor vector based on the count of each level in the data.

factor_vec <- factor(c('apple', 'banana', 'apple', 'cherry', 'banana', 'banana', 'date'))

# Reorder levels by decreasing count
ft_count(factor_vec)
## [1] apple  banana apple  cherry banana banana date  
## Levels: banana apple cherry date
# Reorder levels by increasing count
ft_count(factor_vec, decreasing = FALSE)
## [1] apple  banana apple  cherry banana banana date  
## Levels: cherry date apple banana

ft_sub Reorders the levels of a factor vector based on substrings extracted from the factor levels.

factor_vec <- factor(c('Apple', 'banana', 'Cherry', 'date', 'Fig', 'grape'))
# Reorder based on substring from position 2 to 4
ft_sub(factor_vec, start_pos = 2, end_pos = 4)
## [1] banana date   Cherry Fig    Apple  grape 
## Levels: banana date Cherry Fig Apple grape
# Reorder from position 3 to end, case-sensitive
ft_sub(factor_vec, start_pos = 3, case = TRUE)
## [1] grape  Cherry Fig    banana Apple  date  
## Levels: grape Cherry Fig banana Apple date

ft_freq Reorders the levels of a factor vector based on the total frequency of characters appearing in the vector.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date', 'banana', 'apple', 'fig'))

# Reorder levels based on total character frequency
ft_freq(factor_vec)
## [1] apple  banana cherry date   banana apple  fig   
## Levels: banana apple date cherry fig
# Reorder levels, case-sensitive
factor_vec_case <- factor(c('Apple', 'banana', 'Cherry', 'date', 'banana', 'apple', 'Fig'))
ft_freq(factor_vec_case, case = TRUE)
## [1] Apple  banana Cherry date   banana apple  Fig   
## Levels: banana apple Apple date Cherry Fig

ft_char_freq Reorders the levels of a factor vector based on the frequency of characters at specified positions within the data.

factor_vec <- factor(c('apple', 'banana', 'apricot', 'cherry', 'banana', 'banana', 'date'))

# Reorder based on characters at positions 1 and 2
ft_char_freq(factor_vec, positions = 1:2)
## [1] banana  banana  banana  apricot apple   date    cherry 
## Levels: banana apricot apple date cherry
# Reorder, case-sensitive, decreasing order
ft_char_freq(factor_vec, positions = c(1, 3), case = TRUE)
## [1] banana  banana  banana  date    cherry  apricot apple  
## Levels: banana date cherry apricot apple

ft_substr_freq Reorders the levels of a factor vector based on the frequency of substrings extracted from the data.

factor_vec <- factor(c('apple', 'banana', 'apricot', 'cherry', 'banana', 'banana', 'date'))
ft_substr_freq(factor_vec, start_pos = 2, end_pos=3)
## [1] banana  banana  banana  date    cherry  apricot apple  
## Levels: banana date cherry apricot apple

ft_regex_freq Reorders the levels of a factor vector based on the frequency of substrings matching a regular expression.

factor_vec <- factor(c('apple', 'banana', 'apricot', 'cherry', 'blueberry', 'blackberry', 'date'))

# Reorder based on pattern matching 'a'
ft_regex_freq(factor_vec, pattern = 'a')
## [1] date       blackberry banana     apricot    apple      cherry     blueberry 
## Levels: date blackberry banana apricot apple cherry blueberry
# Reorder with case-sensitive matching
ft_regex_freq(factor_vec, pattern = '^[A-Z]', case = TRUE)
## [1] date       cherry     blueberry  blackberry banana     apricot    apple     
## Levels: date cherry blueberry blackberry banana apricot apple

ft_split Splits the levels of a factor vector using specified patterns or positions and reorders based on specified parts or criteria.

# Example factor vector with patterns
factor_vec <- factor(c('item1-sub1', 'atem2_aub2', 'item3|sub3', 'item1-sub4'))

# Split by patterns '-', '_', or '|' and reorder based on the first part
ft_split(factor_vec, split_pattern = c('-', '_', '\\|'), part = 1)
## [1] item1-sub1 atem2_aub2 item3|sub3 item1-sub4
## Levels: atem2_aub2 item1-sub1 item1-sub4 item3|sub3
# Use the second pattern '_' for splitting
ft_split(factor_vec, split_pattern = c('-', '_', '\\|'), use_pattern = 2, part = 2)
## [1] item1-sub1 atem2_aub2 item3|sub3 item1-sub4
## Levels: item1-sub1 item3|sub3 item1-sub4 atem2_aub2
# Reorder based on character frequencies in the specified part
ft_split(factor_vec, split_pattern = '-', part = 2, char_freq = TRUE)
## [1] item1-sub1 atem2_aub2 item3|sub3 item1-sub4
## Levels: atem2_aub2 item3|sub3 item1-sub1 item1-sub4

ft_len Reorders the levels of a factor vector based on the character length of each level.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date'))

# Sort levels by length
ft_len(factor_vec)
## [1] apple  banana cherry date  
## Levels: date apple banana cherry

ft_sort Sorts the levels of a factor vector based on the values of another vector or a column from a data frame. Handles cases where the sorting vector may contain NAs.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date'))
by_vec <- c(2, 3, 1, NA)
ft_sort(factor_vec, by = by_vec)
## [1] apple  banana cherry date  
## Levels: cherry apple banana date
# Example using a data frame column
data <- data.frame(
  Category = factor(c('apple', 'banana', 'cherry', 'date')),
  Value = c(2, 3, 1, NA)
)
ft_sort(data$Category, by = data$Value)
## [1] apple  banana cherry date  
## Levels: cherry apple banana date

ft_sort_custom Reorders the levels of a factor vector based on a custom function applied to each level.

factor_vec <- factor(c('apple', 'banana', 'cherry'))

# Sort levels by reverse alphabetical order
ft_sort_custom(factor_vec, function(x) -rank(x))
## [1] apple  banana cherry
## Levels: cherry banana apple
# Sort levels by length of the level name
ft_sort_custom(factor_vec, function(x) nchar(x))
## [1] apple  banana cherry
## Levels: apple banana cherry

2.2 Replacing Factor Levels

ft_replace Replaces a specified level in a factor vector with a new level. If a position is provided, the new level is inserted at the specified position among the levels; otherwise, the original level order is preserved.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date', 'fig', 'grape'))

# replace 'banana' as 'blueberry', and keep original order
ft_replace(factor_vec, old_level = 'banana', new_level = 'blueberry')
## [1] apple     blueberry cherry    date      fig       grape    
## Levels: apple blueberry cherry date fig grape
# replace 'banana' as 'blueberry'
ft_replace(factor_vec, old_level = 'banana', new_level = 'blueberry', position = 2)
## [1] apple     blueberry cherry    date      fig       grape    
## Levels: apple blueberry cherry date fig grape

ft_replace_pattern Replaces parts of the factor levels that match a specified pattern with a new string.

factor_vec <- factor(c('apple_pie', 'banana_bread', 'cherry_cake'))

# Replace '_pie', '_bread', '_cake' with '_dessert'
ft_replace_pattern(factor_vec, pattern = '_.*', replacement = '_dessert')
## [1] apple_dessert  banana_dessert cherry_dessert
## Levels: apple_dessert banana_dessert cherry_dessert

2.3 Filtering and Removing Factor Levels

ft_filter_freq Filters out factor levels that occur less than a specified frequency threshold and recalculates character frequencies excluding the removed levels. Offers options to handle NA values and returns additional information.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date', 'banana', 'apple', 'fig', NA))

# Filter levels occurring less than 2 times and reorder by character frequency
ft_filter_freq(factor_vec, min_freq = 2)
## [1] apple  banana banana apple 
## Levels: banana apple
# Filter levels, remove NA values, and return additional information
result <- ft_filter_freq(factor_vec, min_freq = 2, na.rm = TRUE, return_info = TRUE)
result$filtered_factor
## [1] apple  banana banana apple 
## Levels: banana apple
result$removed_levels
## [1] "cherry" "date"   "fig"
result$char_freq_table
## all_chars
## a b e l n p 
## 8 2 2 2 4 4

ft_filter_pos Removes factor levels where a specified character appears at specified positions within the levels.

factor_vec <- factor(c('apple', 'banana', 'apricot', 'cherry', 'date', 'fig', 'grape'))

# Remove levels where 'a' appears at position 1
ft_filter_pos(factor_vec, positions = 1, char = 'a')
## [1] banana cherry date   fig    grape 
## Levels: banana cherry date fig grape
# Remove levels where 'e' appears at positions 2 or 3
ft_filter_pos(factor_vec, positions = c(2, 3), char = 'e')
## [1] apple   banana  apricot date    fig     grape  
## Levels: apple apricot banana date fig grape
# Case-sensitive removal
factor_vec_case <- factor(c('Apple', 'banana', 'Apricot', 'Cherry', 'Date', 'Fig', 'grape'))
ft_filter_pos(factor_vec_case, positions = 1, char = 'A', case = TRUE)
## [1] banana Cherry Date   Fig    grape 
## Levels: Cherry Date Fig banana grape

ft_remove_levels Removes specified levels from a factor vector, keeping the remaining levels and their order unchanged.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date', 'fig', 'grape'))

# Remove levels 'banana' and 'date'
ft_remove_levels(factor_vec, levels_to_remove = c('banana', 'date'))
## [1] apple  cherry fig    grape 
## Levels: apple cherry fig grape

ft_filter_func Removes levels from a factor vector based on a user-defined function.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date'))

# Remove levels that start with 'b'
ft_filter_func(factor_vec, function(x) !grepl('^b', x))
## [1] apple  <NA>   cherry date  
## Levels: apple cherry date

2.4 Merging Factor Vectors

ft_merge_similar Merges levels of a factor that are similar based on string distance.

factor_vec <- factor(c('apple', 'appel', 'banana', 'bananna', 'cherry'))

# Merge similar levels
ft_merge_similar(factor_vec, max_distance = 1)
## [1] apple  appel  banana banana cherry
## Levels: appel apple banana cherry

ft_concat Combines multiple factor vectors into a single factor, unifying the levels.

factor_vec1 <- factor(c('apple', 'banana'))
factor_vec2 <- factor(c('cherry', 'date'))

# Concatenate factors
concatenated_factor <- ft_concat(factor_vec1, factor_vec2)
levels(concatenated_factor)
## [1] "apple"  "banana" "cherry" "date"

ft_combine Combines two vectors, which may be of unequal lengths, into a factor vector and sorts based on the levels of either the first or second vector.

vector1 <- c('apple', 'banana', 'cherry')
vector2 <- c('date', 'fig', 'grape', 'honeydew')

# Combine and sort based on vector1 levels
ft_combine(vector1, vector2, sort_by = 1)
## [1] apple    banana   cherry   date     fig      grape    honeydew
## Levels: apple banana cherry date fig grape honeydew
# Combine and sort based on vector2 levels
ft_combine(vector1, vector2, sort_by = 2)
## [1] apple    banana   cherry   date     fig      grape    honeydew
## Levels: date fig grape honeydew apple banana cherry

2.5 Other Useful Functions

ft_insert Inserts one or more new levels into a factor vector immediately after specified target levels. Targets can be identified by exact matches, positions, or pattern-based matching. Supports case sensitivity and handling of \code{NA} values. Can handle multiple insertions and maintains the original order of other levels. If a new level already exists in the factor and \code{allow_duplicates} is \code{FALSE}, it is moved to the desired position without duplication. If \code{allow_duplicates} is \code{TRUE}, unique duplicates are created.

factor_vec <- factor(c('apple', 'banana', 'cherry', 'date', 'fig', 'grape'))
ft_insert(factor_vec, insert = 'date', target = 'banana', inplace = TRUE)
## [1] apple  banana date   cherry fig    grape 
## Levels: apple banana date cherry fig grape
ft_insert(factor_vec, insert = c('date', 'grape'), positions = c(2, 4))
## [1] apple  banana cherry date   fig    grape 
## Levels: apple banana date cherry grape fig
ft_insert(factor_vec, insert = 'honeydew', pattern = '^c')
## [1] apple  banana cherry date   fig    grape 
## Levels: apple banana cherry honeydew date fig grape
factor_vec_na <- factor(c('apple', NA, 'banana', 'cherry', NA, 'date'))
ft_insert(factor_vec_na, insert = 'lychee', insert_after_na = TRUE)
## Warning in ft_insert(factor_vec_na, insert = "lychee", insert_after_na = TRUE):
## No target levels found for insertion. Returning the original factor.
## [1] apple  <NA>   banana cherry <NA>   date  
## Levels: apple banana cherry date

ft_intersect Combines multiple factor vectors and returns a factor vector containing only the levels common to all.

factor_vec1 <- factor(c('apple', 'banana', 'cherry'))
factor_vec2 <- factor(c('banana', 'date', 'cherry'))
factor_vec3 <- factor(c('banana', 'cherry', 'fig'))

# Get intersection of levels
ft_intersect(factor_vec1, factor_vec2, factor_vec3)
## [1] banana cherry banana cherry banana cherry
## Levels: banana cherry

ft_union Combines multiple factor vectors and returns a factor vector containing all unique levels.

factor_vec1 <- factor(c('apple', 'banana'))
factor_vec2 <- factor(c('banana', 'cherry'))
factor_vec3 <- factor(c('date', 'fig'))

# Get union of levels
ft_union(factor_vec1, factor_vec2, factor_vec3)
## [1] apple  banana banana cherry date   fig   
## Levels: apple banana cherry date fig

ft_reorder_within Reorders the levels of a factor vector within groups defined by another factor vector.

data <- data.frame(
  item = factor(c('A', 'B', 'C', 'D', 'E', 'F')),
  group = factor(c('G1', 'G1', 'G1', 'G2', 'G2', 'G2')),
  value = c(10, 15, 5, 20, 25, 15)
)
data <- rbind(data, data)
# Reorder 'item' within 'group' by 'value'
data$item <- ft_reorder_within(data$item, data$group, data$value, mean)

ft_extract Extracts substrings from the levels of a factor vector based on a regular expression pattern and creates a new factor.

factor_vec <- factor(c('item123', 'item456', 'item789'))

# Extract numeric part
ft_extract(factor_vec, pattern = '\\d+')
## [1] 123 456 789
## Levels: 123 456 789
# Extract with capturing group
factor_vec <- factor(c('apple: red', 'banana: yellow', 'cherry: red'))
ft_extract(factor_vec, pattern = '^(\\w+):', capture_group = 1)
## [1] apple  banana cherry
## Levels: apple banana cherry

ft_pad_levels Pads the levels of a factor vector with leading characters to achieve a specified width.

# Example factor vector
factor_vec <- factor(c('A', 'B', 'C', 'D'))

# Pad levels to width 4 using '0' as padding character
padded_factor <- ft_pad_levels(factor_vec, width = 4, pad_char = '0')
print(levels(padded_factor))
## [1] "000A" "000B" "000C" "000D"
# Output: "000A" "000B" "000C" "000D"

# Pad levels to width 6 using '%A' as padding string
padded_factor <- ft_pad_levels(factor_vec, width = 6, pad_char = '%A')
print(levels(padded_factor))
## [1] "%A%%A%A" "%A%%A%B" "%A%%A%C" "%A%%A%D"
# Output: "%%A%A" "%%A%B" "%%A%C" "%%A%D"

ft_level_stats Computes statistical summaries for each level of a factor vector based on associated numeric data. (group_by and summarize).

ft_pattern_replace Replaces substrings in factor levels that match a pattern with a replacement string.

ft_impute Replaces \code{NA} values in a factor vector using specified imputation methods.

ft_unique_comb Generates a new factor where each level represents a unique combination of levels from the input factors.

ft_map_func Transforms factor levels by applying a function that can include complex logic.

ft_collapse_lev Collapses specified levels of a factor into new levels based on a grouping list.

ft_duplicates Identifies duplicate levels in a factor vector and returns a logical vector indicating which elements are duplicates.

ft_dummy Generates a data frame of dummy variables (one-hot encoded) from a factor vector.

ft_replace_na Replaces \code{NA} values in a factor vector with a specified level.

ft_sample_levels Randomly selects a specified number of levels from a factor vector.

ft_apply Transforms factor levels by applying a function to each level.

ft_encode Converts the levels of a factor vector into numeric codes, optionally using a provided mapping.

3. Summary

The fctutils package provides a comprehensive set of functions to efficiently manage and manipulate factor vectors in R. From ordering and sorting to replacing, filtering, merging, and beyond, these tools enhance your ability to handle categorical data with ease. The additional essential functions further extend the package’s capabilities, ensuring that all common factor operations are covered.

4. Contact information

For any questions please contact guokai8@gmail.com or submit the issues to https://github.com/guokai8/fctutils/issues