logitModel: Alternative Logistic Regression Implementation

In mr-avila/logitModel: Simple Alternative to Logit Modelling in R

Description

logitModel computes a logistic regression for a binary response variable via newtonRaphson estimation method.

Usage

logitModel(formula, data, precision = 1e-10, iterMax = 25)

Arguments

formula	an object of class "formula".
data	an optional data frame or coercible to one. If empty, the variables are taken from the Global Environment.
precision	degree of precision of optimisation algorithm.
iterMax	maximum number of iterations of optimisation algorithm.

Value

logitModel returns an object of class logitModel, which includes the coefficients, fitted vlaues, variance-covariance matrix, residuals, degree of freedom, loglikelihood value and deviance (these last three values also from Null model), and AIC. A few extra extra inforamtion such as the formula, the function call, the response variable and the design matrix, as well as the weight matrix and the number of iterations needed for convergence.

Examples

set.seed(42)
X1 <- rnorm(100);    X2 <- rnorm(100);
points <- 2 * X1 - 3 * X2
y <- rbinom(100, 1, exp(points) / (1 + exp(points)))
X <- cbind(b1 = 1, b2 = X1, b3 = X2)
fit <- logitModel(y ~ X1 + X2)

mr-avila/logitModel documentation built on July 13, 2021, 6:48 p.m.