The bulkreadr
package in R includes specialized
functions beyond bulk data reading, aimed at enhancing data analysis
efficiency. These functions are designed to operate on individual
vectors, except for inspect_na()
and
fill_missing_values()
, which work on data frames.
pull_out()
is similar to [. It acts on vectors,
matrices, arrays and lists to extract or replace parts. It is pleasant
to use with the magrittr (%>%
) and
base(|>
) operators.
library(bulkreadr)
library(dplyr)
top_10_richest_nig <- c("Aliko Dangote", "Mike Adenuga", "Femi Otedola", "Arthur Eze", "Abdulsamad Rabiu", "Cletus Ibeto", "Orji Uzor Kalu", "ABC Orjiakor", "Jimoh Ibrahim", "Tony Elumelu")
top_10_richest_nig %>%
pull_out(c(1, 5, 2))
#> [1] "Aliko Dangote" "Abdulsamad Rabiu" "Mike Adenuga"
convert_to_date()
parses an input vector into POSIXct
date-time object. It is also powerful to convert from excel date number
like 42370
into date value like
2016-01-01
.
## ** heterogeneous dates **
dates <- c(
44869, "22.09.2022", NA, "02/27/92", "01-19-2022",
"13-01- 2022", "2023", "2023-2", 41750.2, 41751.99,
"11 07 2023", "2023-4"
)
# Convert to POSIXct or Date object
convert_to_date(dates)
#> [1] "2022-11-04" "2022-09-22" NA "1992-02-27" "2022-01-19"
#> [6] "2022-01-13" "2023-01-01" "2023-02-01" "2014-04-21" "2014-04-22"
#> [11] "2023-07-11" "2023-04-01"
# It can also convert date time object to date object
convert_to_date(lubridate::now())
#> [1] "2024-05-26"
inspect_na()
summarizes the rate of missingness in each
column of a data frame. For a grouped data frame, the rate of
missingness is summarized separately for each group.
# dataframe summary
inspect_na(airquality)
#> # A tibble: 6 × 3
#> col_name cnt pcnt
#> <chr> <int> <dbl>
#> 1 Ozone 37 24.2
#> 2 Solar.R 7 4.58
#> 3 Wind 0 0
#> 4 Temp 0 0
#> 5 Month 0 0
#> # ℹ 1 more row
Grouped dataframe summary
fill_missing_values()
is an efficient function that
addresses missing values in a data frame. It uses imputation by
function, also known as column-based imputation, to impute the missing
values. It supports various imputation methods for continuous variables,
including minimum
, maximum
, mean
,
median
, harmonic mean
, and
geometric mean
. For categorical variables, missing values
are replaced with the mode
of the column. This approach
ensures accurate and consistent replacements derived from individual
columns, resulting in a complete and reliable dataset for improved
analysis and decision-making.
df <- tibble::tibble(
Sepal_Length = c(5.2, 5, 5.7, NA, 6.2, 6.7, 5.5),
Sepal.Width = c(4.1, 3.6, 3, 3, 2.9, 2.5, 2.4),
Petal_Length = c(1.5, 1.4, 4.2, 1.4, NA, 5.8, 3.7),
Petal_Width = c(NA, 0.2, 1.2, 0.2, 1.3, 1.8, NA),
Species = c("setosa", NA, "versicolor", "setosa",
NA, "virginica", "setosa"
)
)
df
#> # A tibble: 7 × 5
#> Sepal_Length Sepal.Width Petal_Length Petal_Width Species
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 5.2 4.1 1.5 NA setosa
#> 2 5 3.6 1.4 0.2 <NA>
#> 3 5.7 3 4.2 1.2 versicolor
#> 4 NA 3 1.4 0.2 setosa
#> 5 6.2 2.9 NA 1.3 <NA>
#> # ℹ 2 more rows
If you do not specify selected_variables
(i.e., leave it
as NULL
), the function will impute missing values for all
columns in the dataframe.
# Impute using the mean
result_df_mean <- fill_missing_values(df, method = "mean")
result_df_mean
#> # A tibble: 7 × 5
#> Sepal_Length Sepal.Width Petal_Length Petal_Width Species
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 5.2 4.1 1.5 0.94 setosa
#> 2 5 3.6 1.4 0.2 setosa
#> 3 5.7 3 4.2 1.2 versicolor
#> 4 5.72 3 1.4 0.2 setosa
#> 5 6.2 2.9 3 1.3 setosa
#> # ℹ 2 more rows
If you specify column names, only those columns will be imputed. For
example, impute for variables Petal_Length
and
Petal_Width
using the geometric mean.
result_df_geomean <- fill_missing_values(df, selected_variables = c
("Petal_Length", "Petal_Width"), method = "geometric")
result_df_geomean
#> # A tibble: 7 × 5
#> Sepal_Length Sepal.Width Petal_Length Petal_Width Species
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 5.2 4.1 1.5 0.732 setosa
#> 2 5 3.6 1.4 0.2 <NA>
#> 3 5.7 3 4.2 1.2 versicolor
#> 4 NA 3 1.4 0.2 setosa
#> 5 6.2 2.9 2.22 1.3 <NA>
#> # ℹ 2 more rows
If you specify column positions, only the columns at those positions will be imputed.
# Impute using the maximum method
result_df_max <- fill_missing_values(df, selected_variables = c
(2, 3), method = "max")
result_df_geomean
#> # A tibble: 7 × 5
#> Sepal_Length Sepal.Width Petal_Length Petal_Width Species
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 5.2 4.1 1.5 0.732 setosa
#> 2 5 3.6 1.4 0.2 <NA>
#> 3 5.7 3 4.2 1.2 versicolor
#> 4 NA 3 1.4 0.2 setosa
#> 5 6.2 2.9 2.22 1.3 <NA>
#> # ℹ 2 more rows
You can use the fill_missing_values()
in a grouped data
frame by using other grouping and map functions. Here is an example of
how to do this:
sample_iris <- tibble::tibble(
Sepal_Length = c(5.2, 5, 5.7, NA, 6.2, 6.7, 5.5),
Petal_Length = c(1.5, 1.4, 4.2, 1.4, NA, 5.8, 3.7),
Petal_Width = c(0.3, 0.2, 1.2, 0.2, 1.3, 1.8, NA),
Species = c("setosa", "setosa", "versicolor", "setosa",
"virginica", "virginica", "setosa")
)
sample_iris
#> # A tibble: 7 × 4
#> Sepal_Length Petal_Length Petal_Width Species
#> <dbl> <dbl> <dbl> <chr>
#> 1 5.2 1.5 0.3 setosa
#> 2 5 1.4 0.2 setosa
#> 3 5.7 4.2 1.2 versicolor
#> 4 NA 1.4 0.2 setosa
#> 5 6.2 NA 1.3 virginica
#> # ℹ 2 more rows
sample_iris %>%
group_by(Species) %>%
group_split() %>%
map_df(fill_missing_values, method = "median")
#> # A tibble: 7 × 4
#> Sepal_Length Petal_Length Petal_Width Species
#> <dbl> <dbl> <dbl> <chr>
#> 1 5.2 1.5 0.3 setosa
#> 2 5 1.4 0.2 setosa
#> 3 5.2 1.4 0.2 setosa
#> 4 5.5 3.7 0.2 setosa
#> 5 5.7 4.2 1.2 versicolor
#> # ℹ 2 more rows