rem_covars: Convert covariate factors to logical vector

In jmueller17/sociometrics: Managing and analyzing timestamped wearable sensor data

Convert covariate factors to logical vector

Description

Converts a factor (or character) vector into a logical vector (or matrix) where the reference category is set to TRUE and all other levels to FALSE. The resulting logical vector (or matrix) is used as covariate parameter in the relevent::rem.dyad() function.

There are two modes: either retrieves a logical vector or a n x n difference/similarity matix of node attributes.

Usage

rem_covars(
  df_covars,
  ids = NULL,
  covar_v = NULL,
  covar_m = NULL,
  covar_t = "attr",
  idcol = NULL,
  ...
)

Arguments

df_covars	Dataframe or matrix which contains covariate information. This can either be a dataframe where columns specify sociodemographic variables such as age, gender, etc. or it can be a matrix which specifies co-location.
ids	Filter for specific ids in df_covars. If ids=NULL, all entries of the df_covars will be retrieved. If idcol=NULL, assumes that ids match row index; otherwise extra column with ids has to be specified in param idcol.
covar_v	Vector of length two. First entry specifies name of attribute to retrieve. This has to correspond to the column name in df_covars. Second entry specifies reference value set to TRUE. If only first entry is provided original attribute values are returned and not a logical vector. Returns logical *vector*
covar_m	Vector of length two or three. First entry specifies name of the attribute to retrieve (e.g. gender). Second entry specifies comparison to be applied such as usually comparison operators "==", "!=" or operators for numeric values such as "-|+". Returns n x n *matrix*.
covar_t	String. Either of c("attr" | "co.loc" | "rr"). Indicating type of covariate data. Usually this is either a vector or a matrix of data attributes. However, there are further two special types, namely the colocation matrix (covar_t="co.loc") and the round robin formats (covar_t="rr"). Default value is "attr".
idcol	Name or index of the column which should be used to filter for ids. Default is NULL meaning that provided ids correspond to the row index.

Details

The order of the values for the covariate vector will be identical to the order in the df_covar dataframe. It is important that the IDs of the nodes has an ascending order since rem.dyad expects its nodes to be ordered from 1..n in ascending order as well. When converting a sociometric data frame to an edge list to be used with relevent package, the default behavior takes care that original values are replaced in ascending order starting with 1..n.

covar_m retrieves a (difference) matrix from covariate attributes. covar_m=c("Gender", "==", "Men") for example retrieves a matrix where men-men dyad are set to 1 and all other combinations of dyads is set to 0. The same code without a reference category specified covar_m=c("Gender", "==") returns a matrix where homophilios ties (men-men or women-women) is set to 1 and all others 0.

Value

either logical vector or data matrix for given attribute containing difference between two node values.

Matrix

See Also

rem_edge_list,

Examples

#Retrieve a numeric vector of the column "Tenure" in dataframe df.attr for badge ids 2,5,7,9
tenure <- rem_covars(df.attr, ids=c(2,5,7,9), covar_v=c("Tenure"))

#Retrieve a logical vector of "Role" Column, now with "Team leader" as reference category and set to TRUE
isLeader <-  rem_covars(df.attr, ids=ub, covar_v=c("Role", "Team leader"))

#Retrieve a n x n matrix of Gender (where n depends on the length of badge IDs in "ub") of "Gender", indicating
#TRUE if two badges have the same "Gender" or FALSE otherwise.
sameSex <- rem_covars(df.attr, ids=ub, covar_m=c("Gender", "=="))

#Retrieve a n x n matrix of Gender where men-men ties are set to 1 and all others to 0
rem_covars(df.attr, covar_m=c("Gender", "==", "Men"))

#Retrieve difference matrix of factor roles; if Role is factor, uses underlying numeric
#representation to construct difference.
rem_covars(df.attr, covar_m=c("Role", "-"))

jmueller17/sociometrics documentation built on March 20, 2024, 1:04 a.m.