linear_regressor: R6 Class Representing a Linear Regression

In StabilizedRegression: Stabilizing Regression and Variable Selection

R6 Class Representing a Linear Regression

Description

An R6-class for linear regression that is used within the StabilizedRegression framework.

Currently this is the only regression procedure that has been implemented. In order to extend the StabilizedRegression framework to a different regression procedure a custom R6-class with the same structure as this function can be written and used within StabilizedRegression.

Details

Constructer method initializes a linear regression object specifying on which subset of variables S to fit the regression and which type of stability test and prediction score to compute. The methods fit() and predict() can be applied to the object to fit and predict, respectively.

Public fields

estimator

Numeric vector of regression coefficients.

S

Numeric vector specifying the subset of variables to perform regression on.

scores

Numeric vector of fitted stability and prediction scores.

pars

List specifying the stability test via test and prediction score via pred_score.

Methods

Public methods

• linear_regressor$new()

• linear_regressor$fit()

• linear_regressor$predict()

• linear_regressor$clone()

---

Method new()

Create a new linear_regression object.

Usage

linear_regressor$new(
  S = numeric(),
  pars = list(test = "mean", pred_score = c("mse", "mse"))
)

Arguments

S

Subset of variables.

pars

Parameters.

Returns

A new 'linear_regression' object.

---

Method fit()

Fit a 'linear_regression' object on data and computes the stability and prediction scores.

Usage

linear_regressor$fit(X, Y, A, extra = NA)

Arguments

X

Predictor matrix.

Y

response vector.

A

environemnt indicator.

extra

not required (placeholder)

Returns

A fitted 'linear_regression' object.

---

Method predict()

Predict using a fitted 'linear_regression' object.

Usage

linear_regressor$predict(X)

Arguments

X

Predictor matrix on which to predict response.

Returns

Numeric vector of predicted response.

---

Method clone()

The objects of this class are cloneable with this method.

Usage

linear_regressor$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Author(s)

Niklas Pfister

StabilizedRegression documentation built on June 30, 2022, 9:06 a.m.