Residuals for a beezdemand_hurdle fit

Response-scale residuals against the marginal (default) or conditional fitted values. type = "pearson" returns the Part-II standardized residual on the model (log-consumption) scale, (log(y) - mu_i) / sigma_e, where mu_i is the subject-conditional linear predictor (predict(type = "link")) and sigma_e is exp(coef[["logsigma_e"]]); observations with y = 0 have no Part-II residual and are NA. (Before 0.3.0 the raw-scale residual was divided by the log-scale sigma_e, which changed with the consumption unit.)

Usage

# S3 method for class 'beezdemand_hurdle'
residuals(object, type = c("response", "pearson"), marginal = TRUE, ...)

Arguments

object

A beezdemand_hurdle object.

type

One of "response" (default) or "pearson".

marginal

Passed to fitted.beezdemand_hurdle(). Default TRUE.

...

Unused.

Value

Numeric vector of length nobs(object).

See also

fitted.beezdemand_hurdle().