oomglm: Out of memory Generalized Linear model

In blakeboswell/yotta: Out-of-memory and memory efficient Linear Model fitting with support for Inference

Description

Perform memory-efficient generalized linear regression using the AS274 bounded memory QR factorization algorithm.

Usage

oomglm(formula, family = gaussian(), weights = NULL, start = NULL,
  ...)

Arguments

formula	A symbolic description of the model to be fitted of class formula.
family	A family object.
weights	A one-sided, single term formula specifying weights.
start	Starting values for the parameters in the linear predictor.
...	Ignored.

Details

An oomglm model can be in various states of fit depending on the number of seen observations and rounds of IRLS that have been performed. It is important to view the model within the context of: the number of observations processed per round of IRLS (n); the number of IRLS iterations that have been performed (iter); and if the IRLS algorithm has converged (converged).

Value

It is up to the user to know when fitting is complete. Therefore, only basic model characteristics are provided as values. Statistics are available on demand via summary and extractor functions.

converged	Indicates if the IRLS algorithm has converged.
iter	The number of iterations of IRLS performed.
n	The number observations processed per round of IRLS.
df.residual	The residual degrees of freedom.
df.null	The residual degrees of freedom.
formula	the formula object specifying the linear model.
family	a family object describing the error distribution and link function used in the model.
terms	The terms object used.
weights	The weights formula provided to the model.
call	The matched call.

See Also

oomlm()

Examples

# The `oomglm()` function employs Iteratively Weighted Least Squares (IWLS).
# The IWLS iterations are performed by the function `fit()` which
# makes passes over the data until convergence.

# reweight 4 times or until convergence
x <- oomglm(mpg ~ cyl + disp)
x <- fit(x, mtcars, times = 4)

tidy(x)

# To fit in-memory data in chunks, use `oomdata_tbl()`:

y      <- oomglm(mpg ~ cyl + disp)
chunks <- oomdata_tbl(mtcars, chunk_size = 10)
y      <- fit(y, chunks, times = 4)

tidy(y)

blakeboswell/yotta documentation built on May 30, 2019, 11:43 a.m.