MLX-backed generalized linear model — mlxs

Fit generalized linear models using iterative reweighted least squares (IRLS) with MLX providing the heavy lifting for weighted least squares solves. Final convergence is done at double precision on the cpu.

Usage

mlxs_glm(
  formula,
  family = mlxs_gaussian(),
  data,
  subset,
  weights,
  na.action = stats::na.exclude,
  start = NULL,
  control = list(),
  ...
)

Arguments

formula

an object of class "formula" (or one that can be coerced to that class): a symbolic description of the model to be fitted. The details of model specification are given under ‘Details’.

family

A mlxs family object (e.g., mlxs_gaussian(), mlxs_binomial(), mlxs_poisson()). You can use "gaussian" etc.

data

an optional data frame, list or environment (or object coercible by as.data.frame to a data frame) containing the variables in the model. If not found in data, the variables are taken from environment(formula), typically the environment from which glm is called.

subset

an optional vector specifying a subset of observations to be used in the fitting process. (See additional details about how this argument interacts with data-dependent bases in the ‘Details’ below.)

weights

an optional vector of ‘prior weights’ to be used in the fitting process. Should be NULL or a numeric vector.

na.action

A function indicating how missing values should be handled. Defaults to stats::na.exclude() so residuals, fitted values, and training-data predictions are padded back to the original row count.

start

starting values for the parameters in the linear predictor.

control

Optional list of control parameters passed to mlxs_glm_control(). Control parameters can include epsilon, epsilon_f64, maxit and trace.

...

For glm: arguments to be used to form the default control argument if it is not supplied directly.

For weights: further arguments passed to or from other methods.

Value

An object of class c("mlxs_glm", "mlxs_model") containing elements similar to the result of stats::glm(). Unlike stats::glm(), rank-deficient model matrices are rejected rather than fit with aliased coefficients.

Examples

fit <- mlxs_glm(mpg ~ cyl + disp, family = mlxs_gaussian(), data = mtcars)
coef(fit)
#> (Intercept)         cyl        disp 
#> 34.66099167 -1.58727658 -0.02058364