solveWeights: A numeric estimator for Optimal Weight Selection

In DylanSpicker/rcalibration: Implements Generalized Regression Calibration (and Related) Estimators for Measurement Error Correction.

Description

This function iteratively solves the equations to estimate for the weights which minimize the MSE of the linear approximation to the truth. That is, this function returns the estimated weights, when these weights are considered parameters in the BLUP equation.

Usage

solveWeights(W, maxit = 500, epsilon = 1e-10, ...)

Arguments

W	A list of length 'k' containing matrices of error-prone proxy measurements of the covariate. Matrices should all be n (observations) x p (dimension of covariates).
maxit	The maximum number of iterations to run. Defaults to 500.
epsilon	The tolerance at which estimates are deemed to have converged (when all weights change by less than epsilon). Defaults to 1e-10.

Value

A list containing $weights, the optimally computed weights, as well as $M_j, the error-covariance structure matrix.

See Also

[rcalibration::getMj()] which this function calls for the $Mj return, [rcalibration::generalizedRC()] which uses these weights

Examples

solveWeights(W)

DylanSpicker/rcalibration documentation built on March 8, 2020, 10:38 a.m.