Classes: Some classes to simplify making predictions

In grayclhn/oosanalysis-R-library: Pseudo out-of-sample forecasting

Description

This file defines some classes and methods that are used to make some of the predictions in the corresponding paper's empirical exercise (Calhoun, 2011).

Usage

CT(model)
Aggregate(model.list, fn)
## S3 method for class 'CT'
predict(object, newdata,...)
## S3 method for class 'Aggregate'
predict(object, newdata,...)

HasMethod(object, method.name)

Arguments

model	An object with a “predict” method.
model.list	A list of objects with “predict” methods.
fn	A function that can take and aggregate a vector: “mean,” for example.
object	A variable belonging to an S3 class.
newdata	A new data set to use to create the new forecasts.
...	For “predict.CT”, additional arguments to pass to the underlying “predict” method. For “predict.Aggregate”, additional arguments to pass to the individual “predict” methods.
method.name	A character vector giving the names of different S3 methods.

Value

“CT” and “Aggregate” return objects with (S3) classes “CT” and “Aggregate” respectively. The “predict” methods each return a single forecast. “HasMethod” returns a logical vector with the same length as “method.name” indicating whether “object” has each method defined.

Author(s)

Gray Calhoun [e-mail]

Examples

  olddata <- data.frame(y = rnorm(30), x = rnorm(30))
  newdata <- data.frame(y = rnorm(3), x = rnorm(3))

  m1 <- lm(y ~ 1, data = olddata)
  m2 <- lm(y ~ x, data = olddata)

  m3 <- CT(m2)
  m4 <- Aggregate(list(m1, m2, m3), median)

  predict(m3, newdata)
  predict(m4, newdata)

  HasMethod(m1, c("plot", "print", "predict", "median"))

grayclhn/oosanalysis-R-library documentation built on May 17, 2019, 8:33 a.m.