Predictions from a CPB fit

Usage

# S3 method for class 'cpb'
predict(
  object,
  newdata = NULL,
  type = c("response", "rate", "link", "ceiling", "prob"),
  at = NULL,
  offset = NULL,
  se.fit = FALSE,
  interval = c("none", "confidence"),
  level = 0.95,
  ...
)

Arguments

object: A "cpb" object.
newdata: Optional data frame of new covariate profiles; if omitted, the fitted data are used.
type: One of "response" (the exact mean of the fitted distribution; for a zero-truncated fit the conditional mean E(Y | Y >= 1)), "rate" (the CPB rate parameter lambda = exp(x'b), which equals the mean only when lambda/(1-alpha) is an integer), "link" (the linear predictor, log lambda), "ceiling" (the implied ceiling lambda/(1-alpha), the CPB's continuous number of trials; the largest count with positive probability is its ceiling on King's support and its floor on the floor support), or "prob" (the probability that Y equals at).
at: For type = "prob", the count value(s) y whose probability is returned (length 1, or one per row of the prediction data).
offset: Optional offset (log scale) for the newdata rows: a numeric vector or a column name in newdata. By default a fit made with an offset column reads that column from newdata, and one made with an offset vector requires this argument. For newdata = NULL the fit's own offset is used.
se.fit, interval, level: The uncertainty of the predicted quantity at the rows of newdata: se.fit = TRUE adds standard errors, and interval = "confidence" a level confidence interval. They follow the rule of first_difference(): the fit's stored bootstrap draws (a percentile interval, and the draws' standard deviation), otherwise the delta method over the fit's covariance, otherwise none. They need newdata. A row at a fixed-effects unit's own effect, or at its own zero-equation dummy, has none, since the draws do not carry that parameter; a row of a unit the fit has not seen, or newdata without the unit column, enters at the reference unit, which they carry. ame() gives average marginal effects.
...: Unused.

Value

A numeric vector. With interval = "confidence", a matrix with columns fit, lwr and upr; with se.fit = TRUE, a list of fit (the vector or the matrix), se.fit and method (each row's interval source). Without se.fit, the source is the attribute method.

Examples

set.seed(1); x <- rnorm(300)
N <- pmax(round(exp(1.6 + 0.5 * x) / 0.5), 1); y <- rbinom(300, N, 0.5)
fit <- cpb(y ~ x, data = data.frame(y = y, x = x)[y > 0, ], se = "none")
predict(fit, newdata = data.frame(x = 0), type = "ceiling")
#>          
#> 10.22242 
predict(fit, newdata = data.frame(x = c(-1, 1)), type = "prob", at = 5)
#> [1] 0.09402889 0.06443154