Run a brms model containing a covariance matrix
brms_cov_model(
Data,
Response,
FixedEffect,
ID,
RandomEffect = NULL,
Matrix,
Chainset = 1,
Family = "gaussian",
Seed = NULL,
Trials = NA,
PriorSamples = TRUE
)A data frame containing the data - covariates should be centered to a the mean or to a meaningful zero (see Schielzeth H. 2010. Simple means to improve the interpretability of regression coefficients. Methods Ecol Evol. 1:103–113. doi:10.1111/j.2041-210X.2010.00012.x.).
String with the name of the column in Data containing the response variable (e.g. "mass").
String with the name of the column in Data containing the fixed effect variable (e.g. "height"). To add multiple fixed effects, use c() (e.g. c("height", "sex")).
String with the name of the column in Data containing the ID of correlated individuals/groups. The current package version allows the use of a single random effect, so no additional random effects are possible yet.
String with the name of the column in Data containing the random effect variable (e.g. "species"). To add multiple random effects, use c() (e.g. c("species", "date")).
A matrix containing the IDs as row/column names and covariance values (ex. A pedigree relatedness matrix).
Defines the number of iterations. Start with Chainset = 1 and increase as needed until convergence. The value of Chainsetis multiplied by 15000 in warmup, 30000 in iterations and 15 in thin intervale. For quick tests use Chainset = 0 (warmup=10; iter=110; thin=10; chains=2)
String to define the family function in the brms model. Current supported families: "gaussian", "binomial", "poisson".
Numeric and optional. Set a seed in order to repeat the results from the model when running it more than once.
The total number of trials in a binomial model. The number of successes should be imputed on Response.
Logical value that defines if brmsfit will contain the priors used. Default is set to FALSE, which does not includ the priors in the brmsfit.
Returns a brmsfit
if (FALSE) { # \dontrun{
library(tidyverse)
# Create a group (ID) variable
md = tibble::tibble(
ID = as.factor(rep(1:10, each = 100))) %>%
# Create a variables
dplyr::mutate(height = rnorm(1000, mean = 170, sd = 10),
mass = 5 + 0.5 * height + rnorm(1000, mean = 0, sd = 5)) %>%
dplyr::mutate(height = height - mean(height))
# Create a covariance matrix (e.g., relatedness matrix)
cov_matrix = matrix(rnorm(10 * 10), 10, 10)
cov_matrix = cov_matrix %*% t(cov_matrix) # Make it positive semi-definite
rownames(cov_matrix) = colnames(cov_matrix) = 1:10
# Ensure the covariance matrix is symmetric
cov_matrix = (cov_matrix + t(cov_matrix)) / 2
mod = brms_cov_model(Chainset = 3,
Response = "mass",
FixedEffect = "height",
ID = "ID",
Matrix = cov_matrix,
Family = "gaussian",
Data = md,
Seed = 0405)
} # }