Skip to contents

Update a COLSA Model with New Data

Source: R/colsa.R
update.colsa.Rd

This function updates a COLSA model with new data, allowing for dynamic adjustment of the model parameters and basis functions.

Usage

# S3 method for class 'colsa'
update(object, newdata, n_basis = "auto", alpha = 1, nu = 0.2, ...)

Arguments

object

An object of class "colsa" representing the fitted COLSA model.

newdata

A data.frame containing the new data to update the model. It must include the variables specified in the original model formula.

n_basis

Either a numeric value specifying the number of basis functions or "auto" to automatically determine the number of basis functions based on the sample size and scaling parameters. Defaults to "auto", which means the number of basis functions will be determined by n_basis = round(alpha * n_samples^nu).

alpha

A positive numeric value controlling the growth rate of the number of basis functions when n_basis = "auto". Defaults to 1.0.

nu

A positive numeric value controlling the scaling exponent for determining the number of basis functions when n_basis = "auto". Defaults to 0.2.

...

Additional arguments passed to the optimization function.

Value

An updated object of class "colsa" with the following components:

  • logLik: The log-likelihood of the updated model.

  • theta: The updated parameter estimates.

  • hessian: The updated Hessian matrix.

  • n_samples: The updated total number of samples.

  • n_basis: A vector tracking the number of basis functions used at each update step.

  • call: The matched call of the update function.

Details

The function updates the COLSA model by incorporating new data and adjusting the number of basis functions dynamically. It ensures that the model parameters and Hessian matrix are updated accordingly. The optimization process is performed using the nlm method to maximize the log-likelihood of the updated model.