Title: Estimate and Manage Empirical Distributions
Description: Tools to estimate and manage empirical distributions, which should work with survey data. One of the main features is the possibility to create data cubes of estimated statistics, that include all the combinations of the variables of interest (see for example functions dcc5() and dcc6()).
Depends: R (≥ 3.1.2)
Imports: magrittr (≥ 1.5), dplyr (≥ 0.7.4), rlang, utils, stats, tidyr (≥ 0.7.0)
Version: 0.0.6
License: GPL-2
URL: https://gibonet.github.io/distrr, https://github.com/gibonet/distrr
Maintainer: Sandro Petrillo Burri <gibo.gaf@gmail.com>
RoxygenNote: 7.1.1
Encoding: UTF-8
LazyData: yes
Collate: "distrr.R" "invented_data.R" "compat-lazyeval.R" "dplyr_new_wrappers.R" "gibutils.R" "jointfuns.R" "Fhat_conditional.R" "distrr_funs.R" "dcc_new.R" "wq_df.R" "dcc6_fixed.R"
NeedsCompilation: no
Packaged: 2020-07-14 07:53:48 UTC; gibo
Author: Sandro Petrillo Burri [aut, cre]
Repository: CRAN
Date/Publication: 2020-07-14 09:20:07 UTC

Weighted empirical cumulative distribution function (ecdf), conditional on one or more variables

Description

Weighted empirical cumulative distribution function (ecdf), conditional on one or more variables

Usage

Fhat_conditional_(.data, .variables, x, weights)

Arguments

.data

a data frame

.variables

a character vector with one or more column names

x

character vector of length one, with the name of the numeric column whose conditional ecdf has to be estimated

weights

character vector of length one, indicating the name of the positive numeric column of weights, which will be used in the estimation of the conditional ecdf

Value

a data frame, with the variables used to condition, the x variable, and columns wsum (aggregated sum of weights, based on unique values of x) and Fhat (the estimated conditional Fhat). In addition to data frame, the object will be of classes grouped_df, tbl_df and tbl (from package dplyr)

Examples

Fhat_conditional_(mtcars,
                 .variables = c("vs", "am"),
                 x = "mpg",
                 weights = "cyl")

Weighted empirical cumulative distribution function (data frame version)

Description

Weighted empirical cumulative distribution function (data frame version)

Usage

Fhat_df_(.data, x, weights)

Arguments

.data

a data frame

x

name of the numeric column (as character)

weights

name of the weight column (as character)

Value

a data frame with columns: x, wcum and Fhat

Examples

data(invented_wages)
Fhat_df_(invented_wages, "wage", "sample_weights")

Generate all combinations of the elements of a character vector

Description

Generate all combinations of the elements of a character vector

Usage

combn_char(x)

Arguments

x

a character vector

Value

a nested list. A list whose elements are lists containing the character vectors with the combinations of their elements.

Examples

combn_char(c("gender", "sector"))
combn_char(c("gender", "sector", "education"))

Data cube creation (dcc)

Description

Data cube creation (dcc)

Usage

dcc(.data, .variables, .fun = jointfun_, ...)

dcc2(.data, .variables, .fun = jointfun_, order_type = extract_unique2, ...)

dcc5(
  .data,
  .variables,
  .fun = jointfun_,
  .total = "Totale",
  order_type = extract_unique4,
  .all = TRUE,
  ...
)

Arguments

.data

data frame to be processed

.variables

variables to split data frame by, as a character vector (c("var1", "var2")).

.fun

function to apply to each piece (default: jointfun_)

...

additional functions passed to .fun.

order_type

a function like extract_unique or extract_unique2.

.total

character string with the name to give to the subset of data that includes all the observations of a variable (default: "Totale").

.all

logical, indicating if functions' have to be evaluated on the complete dataset.

Value

a data cube, with a column for each cateogorical variable used, and a row for each combination of all the categorical variables' modalities. In addition to all the modalities, each variable will also have a "Total" possibility, which includes all the others. The data cube will contain marginal, conditional and joint empirical distributions...

Examples

data("invented_wages")
str(invented_wages)
tmp <- dcc(.data = invented_wages, 
           .variables = c("gender", "sector"), .fun = jointfun_)
tmp
str(tmp)
tmp2 <- dcc2(.data = invented_wages, 
            .variables = c("gender", "education"), 
            .fun = jointfun_, 
            order_type = extract_unique2)
tmp2
str(tmp2)

# dcc5 works like dcc2, but has an additional optional argument, .total,
# that can be added to give a name to the groups that include all the 
# observations of a variable.
tmp5 <- dcc5(.data = invented_wages, 
            .variables = c("gender", "education"),
            .fun = jointfun_,
            .total = "TOTAL",
            order_type = extract_unique2)
tmp5

Data cube creation

Description

Data cube creation

Usage

dcc6(
  .data,
  .variables,
  .funs_list = list(n = ~dplyr::n()),
  .total = "Totale",
  order_type = extract_unique4,
  .all = TRUE
)

dcc6_fixed(
  .data,
  .variables,
  .funs_list = list(n = ~dplyr::n()),
  .total = "Totale",
  order_type = extract_unique5,
  .all = TRUE,
  fixed_variable = NULL
)

Arguments

.data

data frame to be processed.

.variables

variables to split data frame by, as a character vector (c("var1", "var2")).

.funs_list

a list of function calls in the form of right-hand formula.

.total

character string with the name to give to the subset of data that includes all the observations of a variable (default: "Totale").

order_type

a function like extract_unique or extract_unique2.

.all

logical, indicating if functions have to be evaluated on the complete dataset.

fixed_variable

name of the variable for which you do not want to estimate the total

Examples

dcc6(invented_wages,
     .variables = c("gender", "sector"), 
     .funs_list = list(n = ~dplyr::n()),
     .all = TRUE)
     
dcc6(invented_wages,
     .variables = c("gender", "sector"), 
     .funs_list = list(n = ~dplyr::n()),
     .all = FALSE)

Estimate and manage empirical distributions

Description

Tools to estimate and manage empirical distributions, which should work with survey data. One of the main features is the possibility to create data cubes of estimated statistics, that include all the combinations of the variables of interest (see for example functions dcc5() and dcc6()).

Functions to be used in conjunction with 'dcc' family

Description

Functions to be used in conjunction with 'dcc' family

Usage

extract_unique(df)

extract_unique2(df)

extract_unique3(df)

extract_unique4(df)

extract_unique5(df)

Arguments

df

a data frame

Value

a list whose elements are character vectors of the unique values of each column

Examples

data("invented_wages")
tmp <- extract_unique(df = invented_wages[ , c("gender", "sector")])
tmp
str(tmp)

Invented dataset with wages of men and women.

Description

This dataset has been completely invented, in order to do some examples with the package.

Usage

invented_wages

Format

A data frame (tibble) with 1000 rows and 5 variables:

gender

gender of the worker (men or women)

sector

economic sector where the worker is employed (secondary or tertiary)

education

educational level of the worker (I, II or III)

wage

monthly wage of the worker (in an invented currency)

sample_weights

sampling weights

Details

Every row of the dataset consists in a fake/invented individual worker. For every individual there is his/her gender, the economic sector in which he/she works, his/her level of education and his/her wage. Furthermore there is a column with the sampling weights.

A minimal function which counts the number of observations by groups in a data frame

Description

A minimal function which counts the number of observations by groups in a data frame

Usage

jointfun_(.data, .variables, ...)

Arguments

.data

data frame to be processed

.variables

variables to split data frame by, as a character vector (c("var1", "var2")).

...

additional function calls to be applied on the .data

Value

a data frame, with a column for each cateogrical variable used, and a row for each combination of all the categorical variables' modalities.

Examples

data("invented_wages")
tmp <- jointfun_(.data = invented_wages, .variables = c("gender", "sector"))
tmp
str(tmp)

Keeps only joint distribution (removes '.total').

Description

Removes all the rows where variables have value .total.

Usage

only_joint(.cube, .total = "Totale", .variables = NULL)

Arguments

.cube

a datacube with 'Totale' modalities

.total

modality to eliminate (filter out) (default: "Totale")

.variables

a character vector with the names of the categorical variables

Value

a subset of the data cube with only the combinations of all variables modalities, without the "margins".

Examples

data(invented_wages)
str(invented_wages)

vars <- c("gender", "education")
tmp <- dcc2(.data = invented_wages, 
            .variables = vars, 
            .fun = jointfun_, 
            order_type = extract_unique2)
tmp
str(tmp)
only_joint(tmp, .variables = vars)

# Compare dimensions (number of groups)
dim(tmp)
dim(only_joint(tmp, .variables = vars))

Objects exported from other packages

Description

These objects are imported from other packages. Follow the links below to see their documentation.

magrittr

%>%

Empirical weighted quantile

Description

Empirical weighted quantile

Usage

wq(x, weights, probs = c(0.5))

Arguments

x

A numeric vector

weights

A vector of (positive) sample weights

probs

a numeric vector with the desired quantile levels (default 0.5, the median)

Value

The weighted quantile (a numeric vector)

References

Ferrez, J., Graf, M. (2007). Enquète suisse sur la structure des salaires. Programmes R pour l'intervalle de confiance de la médiane. (Rapport de méthodes No. 338-0045). Neuchâtel: Office fédéral de statistique.

Examples

wq(x = rnorm(100), weights = runif(100))