logistic: Logistic Regression and Prediction

In benbren/some.models: Fits a Multiple (or Simple) Linear or Logistic Regression Model

Description

Logistic Regression and Prediction

Usage

logistic(
  X,
  y,
  n = 1,
  i_max = 100,
  tol = 1e-04,
  to_predict = NULL,
  add_intercept = T
)

Arguments

X	a design matrix - no restrictions, but it should have an intercept column or the results will be wrong.
y	an outcome vector. this should either be 1/0 or it should be the number of success out of corresponding n trials (below)
n	the number of trials ^^. if y is 1/0, this should remain 1 and there is only one trial (either 1 or 0) initialized at 1, as most data will come as 1/0.
i_max	as generalized linear models use an iterative algorithm to estimate the parameter, this is the number of iterations of IRWLS that you want to perform.
tol	the tolerance to hop out of the algorithm.
to_predict	a n optional matrix to have predictions made for - should be same dimensions as X, including the intercept, or there will be an error/wrong inference.
add_intercept	TRUE if your design matrix needs an intercept

Value

coefficients, standard errors, wald statistics, p-values, odds ratios and the fitted probabilities. also returns and optional predictions for a set of test data.

Examples

y = rbinom(100,size = 1,prob = 0.4) # see vignette for n != 1 example
X = matrix(rnorm(1000,mean = 0, sd = 10),100,10)
fit = logistic(X,y)
odds_ratios = fit$or

benbren/some.models documentation built on Nov. 25, 2019, 3:27 p.m.