Predictions from a logreg fit

Returns predictions for new data from a fitted logreg model on the link, probability, or class scale, with optional standard errors.

Usage

# S3 method for class 'logreg'
predict(
  object,
  newdata = NULL,
  type = c("link", "response", "class"),
  se.fit = FALSE,
  threshold = 0.5,
  robust = c("none", "HC0", "HC1", "HC2", "HC3"),
  ...
)

Arguments

object

An object of class "logreg".

newdata

A numeric matrix or data frame of new predictors. NULL (default) returns predictions for the training data.

type

Prediction scale.

• "link": the linear predictor (log-odds).

• "response" (default): the fitted probability.

• "class": a hard 0/1 (or factor) label obtained by thresholding the fitted probability at threshold.

se.fit

If TRUE, the returned object carries an "se.fit" attribute holding the per-row standard error. For type = "link" the SE is on the log-odds scale; for type = "response" it is the delta-method SE of the probability. se.fit is not available for type = "class". For bagged fits the SE is the per-row bag standard deviation of the predicted probability. Default FALSE.

threshold

Probability cut-off for type = "class". Default 0.5.

robust

Heteroscedasticity-consistent covariance used for the standard errors.

• "none" (default): the classical maximum-likelihood covariance.

• "HC0", "HC1", "HC2", "HC3": a sandwich estimator.

...

Currently unused.

Value

A numeric vector of predictions on the requested scale (a factor when type = "class" and the fit was built from a factor response). With se.fit = TRUE the result carries an "se.fit" attribute.

Examples

df <- data.frame(y = as.integer(mtcars$am), mtcars[c("wt", "hp")])
fit <- logreg(y ~ wt + hp, data = df)
predict(fit, df[1:5, ], type = "response")
#>         Mazda RX4     Mazda RX4 Wag        Datsun 710    Hornet 4 Drive 
#>        0.84233554        0.40478253        0.97024082        0.04172803 
#> Hornet Sportabout 
#>        0.06938812 
predict(fit, df[1:5, ], type = "class")
#> [1] 1 0 1 0 0