Available Complete-Case Missing Value (ACCMV)

Code for paper "Handling Nonmonotone Missing Data with Available Complete-Case Missing Value Assumption".

• Paper reference: Handling Nonmonotone Missing Data with Available Complete-Case Missing Value Assumption

• Contact: mathchenggang@gmail.com

single_primary_variabe.R

This R script contains functions when there is a single primary variable.

Functions include:

• single_data_preparation: this function prepares the data and finds the missing patterns in the data.
• single_regression_adjustment: this function implements the regression adjustment estimator.
• single_ipw: this function implements the inverse probability weighting (IPW) estimator.
• single_ipw_sensitivity: this function implements the sensitivity analysis with exponential tilting for IPW estimator.
• single_multiply_robust: this function implements the multiply robust estimator.
• single_bootstrap: this function implements the bootstrap to estimate the confidence intervals for the above estimators.

multiple_primary_variables.R

This R script contains functions when there are multiple primary variable. If not specified, the function below assumes that there are two primary variables.

Functions include:

• multiple_data_preparation: similar to the single_data_preparation function above.
• multiple_ra_average: this function implements the regression adjustment estimator when the outcomes are continuous.
• multiple_ra_indicator: this function implements the regression adjustment estimator when the outcomes are binary.
• multiple_ipw: this function implements the inverse probability weighting (IPW) estimator.
• multiple_ipw_sensitivity: this function implements the sensitivity analysis with exponential tilting for IPW estimator.
• multiple_mr_average: this function implements the multiply robust estimator when the outcomes are continuous.
• multiple_mr_indicator: this function implements the multiply robust estimator when the outcomes are binary.
• multiple_bootstrap: this function implements the bootstrap to estimate the confidence intervals for the above estimators.
• ipw_regression: this function output weights for IPW and the weights can be used for estimating the regression parameters.
• bootstrap_regression: this function implements bootstrap for estimating regression parameters.

diabetes_demo.R

An R script for demonstrating how to use the functions mentioned above. We apply our estimators to the Pima Indians Diabetes Dataset.

diabetes.csv

The Pima Indians Diabetes Dataset. It is downloaded from https://www.kaggle.com/datasets/uciml/pima-indians-diabetes-database. This dataset is to diagnostically predict whether or not a patient has diabetes, based on certain diagnostic measurements included in the dataset.