miceFast

Author

Maciej Nasinski

Originally developed as a study project for the Andrzej Nagorko advanced C++ class, miceFast combines low-level optimizations and an object-oriented design to streamline multiple imputation workflows. miceFast is a high-performance R package designed to expedite the process of data imputation, particularly for large datasets where speed and efficiency are paramount.

One of the most notable highlights of miceFast is its blazing-fast implementation of predictive mean matching (PMM). By leveraging presorting and binary search algorithms, miceFast can significantly outperform many existing solutions in R, including the well-known mice package. Indeed, the author’s contributions to mice helped enhance its underlying C-level optimizations—an achievement that now fuels the robust performance of miceFast itself.

Key Features & Advantages:

  1. Speed and Efficiency
    • Presorting and Binary Search: Achieves one of the fastest PMM procedures available in R.
    • Group-Based Operations: Optimizes performance in scenarios where a dataset must be processed by grouping variables.
    • Single Model Evaluation: Multiple imputations are conducted in a single model pass, eliminating redundant computations.
  2. Object-Oriented Paradigm
    • Flexible API: The miceFast class offers an array of methods to fill missing data using techniques such as linear regression, Bayesian approaches, and discriminant analysis.
    • Enhanced Readability & Modularity: Easily integrate custom or advanced imputation strategies without sacrificing clarity.
  3. Integration with Popular R Tools
    • data.table and dplyr: Work seamlessly with major data manipulation packages in R.
    • Boosted BLAS Support: Encourages Windows, Linux, and macOS users to leverage optimized BLAS libraries for potential 100x speedups.
  4. Extensive Documentation & Benchmarks
    • Vignettes & Tutorials: Learn the details of advanced usage via the comprehensive miceFast Intro Vignette.
    • Performance Validation: Real-world benchmarks and performance metrics are available in the performance_validity.R file within the extdata folder.

Installation:

# From CRAN:
install.packages("miceFast")

# Or the development version from GitHub:
# install.packages("devtools")
devtools::install_github("polkas/miceFast")

Why Choose miceFast?
- Tens of thousands of downloads testify to its growing popularity and trust within the R community.
- Offers an object-oriented approach that allows you to fine-tune and experiment with various imputation models.
- Integrated with performance libraries for maximum speed, making it a go-to solution for large-scale or time-sensitive projects.
- Built upon proven optimizations, including direct enhancements to the widely used mice package.

For more information, visit the miceFast website or consult the in-depth Advanced Usage Vignette. Whether you’re an academic researcher tackling large longitudinal datasets or an industry professional refining machine learning pipelines, miceFast delivers a powerful toolkit to handle missing data with unparalleled speed and flexibility.