Skip to contents

Person location estimation with parallel processing

Source: R/easyRasch.R
RIestThetasOLD2.Rd

Yields about 2-3x speed increase when using 4-8 CPU cores. Requires library(furrr)

Usage

RIestThetasOLD2(
  dfin,
  itemParams,
  model = "PCM",
  method = "WL",
  cpu = 4,
  theta_range = c(-7, 7)
)

Arguments

dfin

Dataframe with response data only (no demographics etc), items as columns

itemParams

Optional item (threshold) location matrix

model

Rasch model to use (use NULL for dichotomous data)

method

Estimation method (defaults to "WL")

cpu

Number of CPUs/cores to utilize (default is 4)

theta_range

Range of theta (person location) values

Details

NOTE: Does not yet work with dichotomous data

Outputs a vector of person locations, one for each row in the dataframe.

Uses thetaEst function from catR package to estimate person locations (thetas) for a dataframe with item data as columns and persons as rows. Defaults to use WL estimation (lower bias than ML, see Warm, 1989) and PCM. See ?thetaEst for options available.