Skip to contents

Log-Likelihood and Log-Likelihood correction

Source: R/get_loglikelihood.R
get_loglikelihood.Rd

A robust function to compute the log-likelihood of a model, as well as individual log-likelihoods (for each observation) whenever possible. Can be used as a replacement for stats::logLik() out of the box, as the returned object is of the same class (and it gives the same results by default).

get_loglikelihood_adjustment() can be used to correct the log-likelihood for models with transformed response variables. The adjustment value can be added to the log-likelihood to get the corrected value. This is done automatically in get_loglikelihood() if check_response = TRUE.

Usage

get_loglikelihood(x, ...)

loglikelihood(x, ...)

get_loglikelihood_adjustment(x)

# S3 method for class 'lm'
get_loglikelihood(
  x,
  estimator = "ML",
  REML = FALSE,
  check_response = FALSE,
  verbose = TRUE,
  ...
)

Arguments

x

A model.

...

Passed down to logLik(), if possible.

estimator

Corresponds to the different estimators for the standard deviation of the errors. If estimator="ML" (default), the scaling is done by n (the biased ML estimator), which is then equivalent to using stats::logLik(). If estimator="OLS", it returns the unbiased OLS estimator. estimator="REML" will give same results as logLik(..., REML=TRUE).

REML

Only for linear models. This argument is present for compatibility with stats::logLik(). Setting it to TRUE will overwrite the estimator argument and is thus equivalent to setting estimator="REML". It will give the same results as stats::logLik(..., REML=TRUE). Note that individual log-likelihoods are not available under REML.

check_response

Logical, if TRUE, checks if the response variable is transformed (like log() or sqrt()), and if so, returns a corrected log-likelihood. To get back to the original scale, the likelihood of the model is multiplied by the Jacobian/derivative of the transformation.

verbose

Toggle warnings and messages.

Value

get_loglikelihood() returns an object of class "logLik", also containing the log-likelihoods for each observation as a per_observation attribute (attributes(get_loglikelihood(x))$per_observation) when possible. The code was partly inspired from the nonnest2 package.

get_loglikelihood_adjustment() returns the adjustment value to be added to the log-likelihood to correct for transformed response variables, or NULL if the adjustment could not be computed.

Examples

x <- lm(Sepal.Length ~ Petal.Width + Species, data = iris)

get_loglikelihood(x, estimator = "ML") # Equivalent to stats::logLik(x)
#> 'log Lik.' -101.0339 (df=5)
get_loglikelihood(x, estimator = "REML") # Equivalent to stats::logLik(x, REML=TRUE)
#> 'log Lik.' -107.0896 (df=5)
get_loglikelihood(x, estimator = "OLS")
#> 'log Lik.' -101.0611 (df=5)