logreg2ph: Sieve maximum likelihood estimator (SMLE) for two-phase...

In sarahlotspeich/logreg2ph_R_only: Implements Sieve Maximum Likelihood Estimation (SMLE) Approach for Two-Phase Logistic Regression

Sieve maximum likelihood estimator (SMLE) for two-phase logistic regression problems This function returns the sieve maximum likelihood estimators (SMLE) for the logistic regression model from Lotspeich et al. (2021)

Description

Sieve maximum likelihood estimator (SMLE) for two-phase logistic regression problems This function returns the sieve maximum likelihood estimators (SMLE) for the logistic regression model from Lotspeich et al. (2021)

Usage

logreg2ph(
  Y_unval = NULL,
  Y_val = NULL,
  X_unval = NULL,
  X_val = NULL,
  C = NULL,
  Validated = NULL,
  Bspline = NULL,
  data,
  theta_pred = NULL,
  gamma_pred = NULL,
  initial_lr_params = "Zero",
  h_N_scale = 1,
  noSE = FALSE,
  TOL = 1e-04,
  MAX_ITER = 1000
)

Arguments

Y_unval	Column names with the unvalidated outcome. If Y_unval is null, the outcome is assumed to be error-free.
Y_val	Column names with the validated outcome.
X_unval	Column name(s) with the unvalidated predictors. If X_unval and X_val are null, all precictors are assumed to be error-free.
X_val	Column name(s) with the validated predictors. If X_unval and X_val are null, all precictors are assumed to be error-free.
C	(Optional) Column name(s) with additional error-free covariates.
Validated	Column name with the validation indicator. The validation indicator can be defined as Validated = 1 or TRUE if the subject was validated and Validated = 0 or FALSE otherwise.
Bspline	Vector of column names containing the B-spline basis functions.
data	A dataframe with one row per subject containing columns: Y_unval, Y_val, X_unval, X_val, C, Validated, and Bspline.
theta_pred	Vector of columns in data that pertain to the predictors in the analysis model.
gamma_pred	Vector of columns in data that pertain to the predictors in the outcome error model.
initial_lr_params	Initial values for parametric model parameters. Choices include (1) "Zero" (non-informative starting values) or (2) "Complete-data" (estimated based on validated subjects only)
h_N_scale	Size of the perturbation used in estimating the standard errors via profile likelihood. If none is supplied, default is h_N_scale = 1.
noSE	Indicator for whether standard errors are desired. Defaults to noSE = FALSE.
TOL	Tolerance between iterations in the EM algorithm used to define convergence.
MAX_ITER	Maximum number of iterations allowed in the EM algorithm.

Value

model_coeff	dataframe with final model coefficients and standard error estimates (where applicable) for the analysis model.
outcome_error_coeff	dataframe with final model coefficients for the outcome error model.
bspline_coeff	dataframe with B-spline coefficients for the covariate error model.
converged	indicator of EM algorithm convergence for parameter estimates.
se_converged	indicator of standard error estimate convergence.
converged_msg	(where applicable) description of non-convergence.
iterations	number of iterations completed by EM algorithm to find parameter estimates.

sarahlotspeich/logreg2ph_R_only documentation built on Jan. 20, 2025, 6:20 p.m.