Skip to contents

Convert Model Function to Model Specification

Source: R/fit.R
as_model_spec.Rd

Converts a model function (e.g., lm, glm) to a model specification list with fit and predict functions compatible with the chronofeat API.

Usage

as_model_spec(model_fn, ...)

Arguments

model_fn

A model function that accepts a formula and data argument

...

Additional arguments passed to the model function during fitting

Value

A list with fit and predict functions:

  • fit(y, X, ...) - Fits the model with response y and predictors X

  • predict(object, newdata, ...) - Generates predictions from fitted model

Details

This function provides backward compatibility for users who want to pass model functions directly (e.g., model = lm) instead of model specifications.

The wrapper creates a formula-based interface internally, combining y and X into a temporary data frame and fitting the model using the standard R formula interface.

Supported model functions include any that accept the standard formula/data interface: lm, glm, randomForest::randomForest, ranger::ranger, etc.

Examples

if (FALSE) { # \dontrun{
# Convert lm to model spec
lm_spec <- as_model_spec(lm)

# Convert glm with family argument
glm_spec <- as_model_spec(glm, family = poisson())

# Use directly in fit()
m <- fit(value ~ p(12), data = df, date = "date", model = lm)
} # }