Fit an experimental pre-pilot item-parameter seeding model — fit_item_parameter_seed

Estimates screening predictions for item difficulty/discrimination from item design/process features. Predictions are not calibrated operational parameters.

Usage

fit_item_parameter_seed_model(
  item_data,
  difficulty = "irt_difficulty",
  discrimination = "irt_discrimination",
  predictors,
  engine = c("auto", "ranger", "lm"),
  seed = 2221
)

Arguments

item_data: Calibrated item-level training data.
difficulty, discrimination: Target columns.
predictors: Design/process predictors.
engine: `auto`, `ranger`, or `lm`.
seed: Seed.

Value

An object of class "eye_item_parameter_seed", stored as a named list, with components "difficulty_model", "discrimination_model", "difficulty", "discrimination", "predictors", "engine", "training_data", "status", "caveat". It contains an experimental pre-pilot item-parameter seeding model and associated metadata or diagnostics needed to interpret the result.