mnar

PyGrinder: a Python toolkit for grinding data beans into the incomplete for real-world data simulation by introducing missing values with different missingness patterns, including MCAR (complete at random), MAR (at random), MNAR (not at random), sub sequence missing, and block missing

Time-series missingness simulation separating mechanisms (MCAR/MAR/MNAR) from patterns (pointwise/block/monotone/decay/markov). Weighted multi-driver MAR, exact rate control, automatic calibration, 2D/3D support. Built for reproducible imputation benchmarking.

Source code used for the results reported in the TOIS2021 journal paper.

A library for synthetic missing data generation.

How Different Types of Missingness affect a complete Dataset

Model-based clustering with variable selection for missing data

Healthcare missing-data analysis comparing complete-case, KNN, Bayesian iterative, and multiple imputation under MCAR, MAR, and MNAR.

A R package for dealing with MNAR data.

Fed-CA: complementarity-aware federated aggregation for deep time-series imputation (SAITS) under heterogeneous MNAR, validated on VitalDB clinical data.

Missingness-aware abstention for selective classification under MCAR/MAR/MNAR label missingness.

Benchmark of 11 well-log imputation methods under realistic multi-log missingness — leakage-controlled leave-one-well-out evaluation and reproducibility package.

A comparative study of missing data imputation methods across MCAR, MAR, and MNAR mechanisms using classical, probabilistic, and deep learning approaches.

Additional imputation function for the `R` package `mice`, according to the Heckman model.

A Noise-Resilient Hybrid Imputation-Ensemble (NR-HIE) framework designed to bridge the generalizability gap in medical AI. Utilizing a triple-stream imputation strategy and stacked generalization, this model achieved 81.62% accuracy on external validation data, ensuring robust and medically safe diabetes prediction

FAI (Feature-Wise Adaptive Imputation) is a machine learning framework that automatically selects the best imputation method per feature based on statistical properties — optimizing for downstream predictive performance, not just imputation error.