Construct multiple-response objects

Description

Constructs mr objects representing multiple-choice questions where more than one choice is allowed.

Usage

as.mr(x, ...)
## S3 method for class 'logical'
as.mr(x,name,...)
## S3 method for class 'list'
as.mr(x, sort.levels=TRUE,...,levels=NULL)
## S3 method for class 'factor'
as.mr(x, sort.levels=FALSE,...)
## S3 method for class 'data.frame'
as.mr(x, sort.levels=FALSE,...,na.rm=TRUE)
## S3 method for class 'character'
as.mr(x, sep=", ", sort.levels=TRUE,..., levels=NULL)
## Default S3 method:
as.mr(x, sort.levels=TRUE, levels=unique(x),...)
## S3 method for class 'ms'
as.mr(x,...)

Arguments

Details

The internal representation of mr objects is as a logical matrix with the levels as column names.

The method for logical x coerces a single vector to a one-column matrix, and then applies the name argument as the column name. Given a matrix, the name argument is optional and replaces the existing column names

The method for list x takes a list of character vectors that represent the levels present for one observation. The method for strings splits the string at the supplied separator and then uses the list method.

The method for factor x produces an mr object with the factor levels as levels. Each observation will have only one value.

The data.frame object works for logical or numeric columns of a data frame. Zero or negative values are treated as 'not present', positive values as 'present'. Optionally, NA values are coded as 'not present', which is useful when the data frame was created by reshape or dplyr::spread.

The method for ms objects simply drops the score/rank information

Value

Object of class mr

Examples

nzbirds_list<-list(c("kea","tui"), c("kea","ruru","kaki"), c("ruru"),
c("tui","ruru"), c("tui","kea","ruru"), c("tauhou","kea"))
nzbirds_list
as.mr(nzbirds_list)
as.mr(c("kea, tui","kea, ruru, kaki","ruru","tui, ruru"))

data(nzbirds)
nzbirds
as.mr(nzbirds)

data(ethnicity)
ethnicity
as.logical(ethnicity)
as.mr(as.logical(ethnicity))

Construct scored or ranked multiple-response objects

Description

The internal representation is as a numeric matrix with 0 when a level is not present and the non-zero rank or score when it is present. The data.frame and matrix methods uses the numeric values of x, and by default set NA values to 'not present'. The list method takes a list with a character vector for each observation and uses the position in the list as the rank/score. The character method splits the string at the separators to give a list and uses the list method. The mr method uses a score of 1 whenever the level is present.

Usage

as.ms(x, ...)
## S3 method for class 'list'
as.ms(x,...,levels=NULL)
## S3 method for class 'data.frame'
as.ms(x,...,na.rm=TRUE)
## S3 method for class 'matrix'
as.ms(x,...,na.rm=TRUE)
## S3 method for class 'mr'
as.ms(x,...)
## S3 method for class 'character'
as.ms(x,sep=", ", ...,levels=NULL)

Arguments

Value

Object of class ms

Examples

nzbirds_list<-list(c("kea","tui"), c("kea","ruru","kaki"), c("ruru"),
c("tui","ruru"), c("tui","kea","ruru"), c("tauhou","kea"))
nzbirds_list
(msbirds<-as.ms(nzbirds_list))

(bird_mat <- unclass(msbirds))
as.ms(bird_mat)

Tidyversatile multiple-response objects

Description

The vmr class wraps the mr class using the vctrs package, for compatibility with tidyverse tbl_df objects (tibbles).

Usage

as.vmr(x, ...)
new_vmr(x, levels = unique(do.call(c, x))) 

Arguments

Details

These objects need the vctrs and pillar packages to work, and need the tibble package to be useful.

Value

An object of class vmr

See Also

as.mr

The internals vignette for internal structure

Examples

if (requireNamespace("vctrs", quietly=TRUE)){
   data(nzbirds)
   nzbirds
   tidybirds<-as.vmr(nzbirds, na.rm=TRUE)
   tidybirds
}

Subset of the Great Backyard Bird survey

Description

Counts of observations for 12 bird species by US county and Canadian province in the Great Backyard Bird survey. These birds were randomly sampled from the much larger number in the full data set. See the vignette for more details.

Usage

data("birds")

Format

A data frame with 3046 observations on the following 13 variables.

‘⁠Phalaenoptilus nuttallii⁠’

a numeric vector

‘⁠Fregata magnificens⁠’

a numeric vector

‘⁠Melanerpes lewis⁠’

a numeric vector

‘⁠Melospiza georgiana⁠’

a numeric vector

‘⁠Rallus limicola⁠’

a numeric vector

‘⁠Myioborus pictus⁠’

a numeric vector

‘⁠Poecile gambeli⁠’

a numeric vector

‘⁠Aythya collaris⁠’

a numeric vector

‘⁠Xanthocephalus xanthocephalus⁠’

a numeric vector

‘⁠Gracula religiosa⁠’

a numeric vector

‘⁠Icterus parisorum⁠’

a numeric vector

‘⁠Coccyzus erythropthalmus⁠’

a numeric vector

location

a character vector

Examples

data(birds)
birds<-as.ms(birds[,1:12],na.rm=TRUE)
mtable(as.mr(birds))

Toy example using New Zealand level 1 ethnicity values

Description

The statistical standard for collecting ethnicity data requires that respondents can mark all that are applicable. The level 1 values are "Māori", "Pacific Peoples" (ie, Pacific Island ethnicities), "Asian", "European", and "MELAA" (Middle Eastern, Latin American, and African). This is artificial data

Usage

data("ethnicity")

Format

An object of class mr

Examples

data(ethnicity)
ethnicity

Utility functions for multiple-response objects

Description

These perform diverse useful tasks. mr_count counts the number of levels present for each individual. mr_na sets NA values to something else, ms_na sets them to 0 (ie, not present), mr_drop and ms_drop drop some levels from the object.

Usage

mr_count(x, na.rm = TRUE)
mr_drop(x, levels,...)
ms_drop(x, levels)
mr_na(x, na=TRUE)
ms_na(x)

Arguments

Value

An integer vector for mr_count, an object of class mr, or ms for the other two functions

Examples

data(usethnicity)
race<-as.mr(strsplit(as.character(usethnicity$Q5),""))
mtable(race)
race<-mr_drop(race,c(" ","F","G","H"))
mtable(race)

## to keep just specified levels use [
mtable(race[,c("A","D")])

## How many do people identify with
table(mr_count(race))

data(nzbirds)
seenbirds<-as.mr(nzbirds>0)
countbirds<-mr_count(seenbirds)

## How many types of birds were seen
table(countbirds)

data(ethnicity)
ethnicity
mr_na(ethnicity, FALSE)

Flatten a multiple-response object into a factor

Description

Convert a multiple-response object into a factor using a supplied ordering. Each observation is assigned its first level in the ordering. That is, an observation that has priorities[1] as one of its levels is assigned that value. An observation that does not priorities[1] as one of its levels, but does have priorities[2] is assigned priorities[2].

Usage

mr_flatten(x, priorities, sort=FALSE)

Arguments

Value

A factor

Examples

data(ethnicity)
ethnicity

## NZ 'prioritised ethnicity'
priority<-c("Maori", "Pacific", "Asian", "European/Other")
eth <- mr_na(mr_recode(ethnicity, `European/Other`="European", `European/Other` = "MELAA"), FALSE)

mr_flatten(eth, priority)
mr_flatten(eth, priority, sort=TRUE)

Reorder levels of multiple-response objects

Description

mr_inorder and ms_inorder use the order in which the levels first appear in the data (which is invariant to locale),mr_inseq and ms_inseq sort alphabetically (for the current locale). mr_infreq sorts by frequency, and ms_inscore applies a function to the values in each level – one such function is mean0, which takes the mean of non-zero values. Finally, ms_reorder and mr_reorder use some function of a second variable computed on the observations where each level is present.

Usage

mr_inorder(x,...)
ms_inorder(x)
mr_inseq(x,...)
ms_inseq(x)
mr_infreq(x,na.rm=TRUE,...)
ms_infreq(x)
ms_inscore(x, fun=mean0)
mean0(y)
mr_reorder(x, v, fun=median,...)
ms_reorder(x, v, fun=median)

Arguments

Value

Object of class mr

References

These are based on the reordering functions for factors in the forcats package.

Examples

data(ethnicity)
mr_infreq(ethnicity)
mr_inseq(ethnicity)

data(nzbirds)
mtable(nzbirds)
mtable(ms_inorder(nzbirds))
mtable(ms_inseq(nzbirds))
mtable(ms_inscore(nzbirds, mean0))

Collapse common or rare levels

Description

Combine the least common or most common levels of a mr object into an "other" level.

Usage

mr_lump(x, n, prop, other_level = "Other",
ties.method = c("min", "average", "first", "last", "random", "max"),...)

Arguments

Value

An object of class mr

References

Based on fct_lump from the forcats package.

Examples

data(ethnicity)
mtable(ethnicity)
mtable(mr_lump(ethnicity,2))
mtable(mr_lump(ethnicity,-2))

data(rstudiosurvey)
## Other software being used
other_software<- as.mr(rstudiosurvey[[40]])
mtable(other_software)
## The top 20 responses
common<-mr_lump(other_software, n=20)
mtable(common)
## 'None' isn't really another package
mtable(mr_drop(common,"None"))

## Packages with at least 20% use
mtable(mr_lump(other_software, prop=0.2))

Relabel levels of multiple-response objects

Description

Relabel some or all of the levels of a multiple-response object. Two levels that are recoded to the same value will be combined.

Usage

mr_recode(x, ...)

Arguments

Value

New object of class mr, ms

Examples

data(nzbirds)
nzbirds<-as.mr(nzbirds)
nzbirds
## recode to English names
mr_recode(nzbirds,morepork="ruru",stilt="kaki",waxeye="tauhou")

data(usethnicity)
race<-as.mr(usethnicity$Q5,"")
race<-mr_drop(race,c(" ","F","G","H"))
race <- mr_recode(race, AmIndian="A",Asian="B", Black="C", Pacific="D", White="E")
mtable(race)

Pivot a multiple-response object to long form

Description

Creates a data frame where every observation has as many rows as it has levels present, plus an id column to specify which rows go together.

Usage

mr_stack(x, ..., na.rm = FALSE)
ms_stack(x, ..., na.rm = FALSE)

Arguments

Value

A data frame with columns values and id, plus a column scores if x is a ms object. When more than one object is supplied, the result is an outer join of the two indindividual results, so it contains a row for every combination of an observed value from each object.

Examples

data(ethnicity)
ethnicity
mr_stack(ethnicity)

data(nzbirds)
nzbirds
ms_stack(nzbirds)

## not actually a sensible use
d <- mr_stack(ethnicity, nzbirds)
head(d)
with(d, table(ethnicity, nzbirds))
## equivalent, but more efficient
mtable(mr_na(ethnicity), mr_na(nzbirds))

Set operations on multiple-response objects

Description

These functions take union, intersection, and difference of two multiple-response objects. An observation has a level in the union if it has that level in either input. It has the level in the intersection if it has the level in both inputs. It has the level in the difference if it has the level in x and not in y

Usage

mr_union(x, y,...)
mr_intersect(x, y,...)
mr_diff(x, y,...)

Arguments

Value

Object of class mr

Examples

data(usethnicity)
race<-as.mr(usethnicity$Q5,"")
race<-mr_drop(race,c(" ","F","G","H"))
race <- mr_recode(race, AmIndian="A",Asian="B", Black="C", Pacific="D", White="E")
mtable(race)

hispanic<-as.mr(usethnicity$Q4==1, "Hispanic")
ethnicity<-mr_union(race, hispanic)
mtable(ethnicity)
ethnicity[101:120]

Check if a level or levels is present

Description

Returns vector of TRUE or FALSE according to whether y is onle of the levels present for that row or is the only level present for that row.

Usage

x %has% y
x %hasonly% y
x %hasall% ys
x %hasany% ys

Arguments

Value

Logical vector

Examples

data(ethnicity)

ethnicity
ethnicity %has% "Maori"
ethnicity %hasonly% "Maori"

data(nzbirds)
as.mr(nzbirds) 
as.mr(nzbirds) 

Flatten a scored multiple-response object into a factor

Description

Convert a multiple-response object into a named numeric vector using a supplied ordering.

Usage

ms_flatten(x, priorities, fun, start=0)

Arguments

Details

Each observation is initially assigned the value start. Starting with the lowest-priority level, the current value is combined with the new value as fun(new, current). Using fun=function(x,y) x would return the value for the highest-priority level present; using fun=pmax would return the highest score for any level present; using fun="+" would return the sum of the scores.

Value

A factor

Examples

data(ethnicity)
ethnicity

## NZ 'prioritised ethnicity'
eth <- mr_recode(ethnicity, `European/Other`="European", `European/Other` = "MELAA")
mr_flatten(ethnicity, c("Maori","Pacific","Asian","European/Other"))

data(nzbirds)
## hardest to see first
ms_flatten(nzbirds, c("kaki","ruru","kea","tui","tauhou"),"+")
ms_flatten(nzbirds, c("kaki","ruru","kea","tui","tauhou"),
  fun=function(x,y) x)
ms_flatten(nzbirds, c("kaki","ruru","kea","tui","tauhou"),pmin,start=Inf)

Tables involving multiple-response objects

Description

Creates one-way and two-way tables using every level of a multiple response object. Use table(as.character(x)) to tabulate combinations of levels

Usage

mtable(x, y, na.rm = TRUE)

Arguments

Value

A 1-d or 2-d array with names giving the levels

Examples

data(ethnicity)
mtable(ethnicity)
table(as.character(ethnicity))

data(nzbirds)
nzbirds<-as.mr(nzbirds)

## co-occurence table
mtable(nzbirds, nzbirds)

## table by a factor
v<-rep(c("A","B"),3)
mtable(nzbirds,v)

data(nzbirds)
mtable(nzbirds>0)

Toy example using New Zealand birds

Description

A small artifical dataset that could be produced by asking people to name New Zealand birds. Each observation has scores from 1 (first bird named) to at most 4 (fourth bird named).

Usage

data("nzbirds")

Format

A ms object with 6 observations on the following 5 variables.

kea

a numeric vector

ruru

a numeric vector

tui

a numeric vector

tauhou

a numeric vector

kaki

a numeric vector

Examples

data(nzbirds)
nzbirds
as.mr(nzbirds)

Plot multiple-response objects

Description

The plot method for mr objects is an UpSet plot, showing co-occurences of the various categories. The image method is a heatmap of the variable plotted against itself with mtable.

Usage

## S3 method for class 'mr'
plot(x, ...)
## S3 method for class 'mr'
image(x, type = c("overlap", "conditional", "association",  "raw"), ...) 

Arguments

Value

Used for its side effect

See Also

as.mr, mtable

Examples

data(rstudiosurvey)
other_software<- as.mr(rstudiosurvey[[40]])
## only those with at least 20 responses
common<-mr_lump(other_software, n=20)
common<-mr_drop(common, "None")

## UpSet plot
plot(common)

## images

image(common, type="conditional")
image(common, type="association")

Subset of RStudio 2019 Community Survey

Description

The 'rstudiosurvey' data set contains 1838 rows of responses from the 2019 RStudio Community Survey, where columns are the 51 questions and a column for the timestamp. The variable names are the full questions. Multiple responses are separated by a comma and space. Non-ASCII characters have been converted with the "ASCII//TRANSLIT" option of iconv.

Usage

data("rstudiosurvey")

Format

A data frame with 1838 observations on the following 52 variables.

Timestamp

a character vector

‘⁠How would you rate your level of experience using R?⁠’

a character vector

‘⁠Compared with other technical topics you've learned in school and on the job, on a scale of 1 to 5, how difficult do you expect learning R to be?⁠’

a numeric vector

‘⁠From what you know about R, how long do you expect that it will take for you to learn enough to use R productively?⁠’

a character vector

‘⁠How do you think you would go about the process of learning R?⁠’

a character vector

‘⁠Which statement most closely reflects the primary reason why you are interested in learning R?⁠’

a character vector

‘⁠If you were to learn R, what would do you think you would use it for? (check all that apply)⁠’

a character vector

‘⁠Which analytical tools do you use today for the functions that you might learn R for? (please check all that apply)⁠’

a character vector

‘⁠What do you think is the biggest obstacle you must overcome in trying to learn R? The choices below are only suggestions; if we haven't listed your obstacle, please choose "Other" and add your obstacle in the text. ⁠’

a character vector

‘⁠What year did you first start learning R?⁠’

a numeric vector

‘⁠How did you learn R? If you used multiple methods, please select the one you used the most.⁠’

a character vector

‘⁠Compared with other technical topics you've learned in school and on the job, on a scale of 1 to 5, how difficult has it been for you to learn R?⁠’

a numeric vector

‘⁠Roughly how long did it take you to achieve proficiency in R?⁠’

a character vector

‘⁠Which statement most closely reflects the primary reason why you learned R?⁠’

a character vector

‘⁠What do you think was the biggest obstacle you had to overcome in learning R? The choices below are only suggestions; if we haven't listed your obstacle, please choose "Other" and add your obstacle in the text. ⁠’

a character vector

‘⁠How often do you use R today, either for professional or personal projects?⁠’

a character vector

‘⁠What applications do you use R for most? (check all that apply)⁠’

a character vector

‘⁠Please rate how much you enjoy using R on a scale of 1 to 5, where 1 is you don't enjoy it at all, and 5 is that you enjoy it a great deal.⁠’

a numeric vector

‘⁠How likely are you to recommend R to a colleague, friend, or family member?⁠’

a numeric vector

‘⁠Which tools do you use with your R applications? (please check all that apply)⁠’

a character vector

‘⁠Did you use tidyverse packages such as ggplot2 or dplyr to learn R?⁠’

a character vector

‘⁠Do you use tidyverse packages when you use R now?⁠’

a character vector

‘⁠What do you like best about using R?⁠’

a character vector

‘⁠What do you like least about using R?⁠’

a character vector

‘⁠When you have problems in R, where do you go for help?⁠’

a character vector

‘⁠How do you discover new packages or packages that are unfamiliar to you?⁠’

a character vector

‘⁠How do you share the results that you create in R? Check all that apply.⁠’

a character vector

‘⁠Looking ahead, how do you expect your use of R to change in 2020?⁠’

a character vector

‘⁠To help us ensure that you are not a robot, please enter the number of characters in the word "analysis" in the text box below. Please type your answer as a word; for example if you want 3 to be your answer, type "three".⁠’

a character vector

‘⁠Do you currently use R Markdown? Choose the statement that most closely matches your use.⁠’

a character vector

‘⁠What applications do you use R Markdown for? Check all that apply.⁠’

a character vector

‘⁠Looking forward, how do you expect your use of R Markdown to change in 2020?⁠’

a character vector

‘⁠How often do you currently use Shiny? Choose the statement that most closely matches your use.⁠’

a character vector

‘⁠Looking forward, how do you expect your use of Shiny to change in 2020?⁠’

a character vector

‘⁠Do you currently use Python? Choose the statement that most closely matches your use.⁠’

a character vector

‘⁠What applications do you use Python for most? (check all that apply)⁠’

a character vector

‘⁠Please rate how much you enjoy using Python on a scale of 1 to 5, where 1 is you don't enjoy it at all, and 5 is that you enjoy it a great deal.⁠’

a numeric vector

‘⁠How likely are you to recommend Python to a colleague, friend, or family member?⁠’

a numeric vector

‘⁠Looking forward, how do you expect your use of Python to change in 2020?⁠’

a character vector

‘⁠What computer tools and/or languages have you used besides R?⁠’

a character vector

‘⁠What was the FIRST computer language or tool that you learned?⁠’

a character vector

‘⁠What year were you born?⁠’

a numeric vector

‘⁠What gender do you identify with?⁠’

a character vector

‘⁠I identify my ethnicity as (select all that apply):⁠’

a character vector

‘⁠What is the highest degree or level of school you have completed? If currently enrolled, please use the highest degree received.⁠’

a character vector

‘⁠In what country do you currently reside?⁠’

a character vector

‘⁠What industry do you work or participate in?⁠’

a character vector

‘⁠What is your job title, if any?⁠’

a character vector

‘⁠Which category best describes the work you do?⁠’

a character vector

‘⁠How many people in your organization or work group do you feel that you can ask for help or support when working with R?⁠’

a numeric vector

‘⁠Which of the following events have you attended, if any? Check all that apply.⁠’

a character vector

‘⁠How did you hear about this survey?⁠’

a character vector

Source

https://github.com/rstudio/r-community-survey/tree/master/2019

Examples

data(rstudiosurvey)
names(rstudiosurvey)[40]
## Other software being used
other_software<- as.mr(rstudiosurvey[[40]])
mtable(other_software)
## top 20 responses
common<-mr_lump(other_software, n=20)
mtable(common)
## 'None' isn't really another package
common<-mr_drop(common, "None")
mtable(common)

## UpSet plot
plot(common)

## Excel users filled in the survey later 
timestamp<-as.Date(rstudiosurvey[[1]],format="%m/%d/%y")
boxplot(timestamp~I(common %has% "Excel"))



## names in order of popularity
t<-mtable(common)
popular<-colnames(t)[order(t,decreasing=TRUE)]
## most popular package for each user
cuml_users <- mr_flatten(common, popular, sort=TRUE)
class(cuml_users)
table(cuml_users)

## two-way tables
## people who also use Stata or Julia are less happy with R than those who don't
names(rstudiosurvey)[18]
happy<-factor(rstudiosurvey[[18]])
mtable(happy, common)
round(prop.table(mtable(happy,common),2),2)

## mr objects can be dataframe columns, or expanded to individual levels
df<-data.frame(timestamp, happy, common)
dim(df)
head(df)
df_raw<-data.frame(timestamp, happy, as.matrix(common))
dim(df_raw)
head(df_raw)

Data from Youth Risk Behaviour Survey

Description

This data set contains variables on race and ethnic identification from the 2017 Youth Risk Behaviour Survey, together with two variables on smoking behaviour. The YRBS is a multistage cluster-sampled survey, so valid inference about associations requires using survey design information. This subset is useful only for demonstration purposes.

Usage

data("usethnicity")

Format

A data frame with 14765 observations on the following 4 variables.

Q4

1 is "Hispanic or Latino

Q5

Character string with zero or more of: A. American Indian or Alaska Native, B. Asian, C. Black or African American, D. Native Hawaiian or Other Pacific Islander, E. White

QN30

1 is "smoked cigarettes on one or more of the past 30 days"

QN31

1 is "smoked more than 10 cigarettes per day on the days they smoked during the past 30 days", those who did not smoke at all are NA

Source

https://www.cdc.gov/healthyyouth/data/yrbs/data.htm

Examples

data(usethnicity)
race<-as.mr(strsplit(as.character(usethnicity$Q5),""))
race<-mr_drop(race," ")
mtable(race)
hispanic<-as.mr(usethnicity$Q4==1,"Hispanic")
ethnicity<-mr_union(race,hispanic)
ethnicity[101:120]