dh.multGLM: Loop multiple GLM models and handle errors & non-convergence

In lifecycle-project/ds-cs-functions: Helper Functions for Use with 'DataSHIELD'

Loop multiple GLM models and handle errors & non-convergence

Description

This function allows you to fit different glm models to different cohorts, for example if you want to include different covariates for different cohorts.

Usage

dh.multGLM(
  df = NULL,
  ref = NULL,
  checks = TRUE,
  conns = NULL,
  vary_df = F,
  family = "gaussian",
  weights = NULL
)

Arguments

df	Character specifying a server-side data frame.
ref	reference tibble, output from dh.buildModels.
checks	Logical; if TRUE checks are performed prior to running the function. Default is TRUE.
conns	DataSHIELD connections object.
vary_df	Option to provide different df for different models. Default is FALSE.
family	Family to use in glm models. Default is "gaussian".
weights	Optional, serverside object containing weights to be used in the model.

Value

Tibble containing five columns:

• model = Description of model fit, taken from model_name.

• formula = Formula for this model, taken from formulae.

• cohort = Cohort for which this model was fit, taken from cohort.

• fit = Output from glm model.

• converged = Logical; whether or not the model converged/fit'.

lifecycle-project/ds-cs-functions documentation built on Nov. 18, 2024, 3:36 p.m.