Ultra data-oriented Parallel FHDI (UP-FHDI)
The version 1.0 of the ultra data-oriented parallel fractional hot deck imputation (UP-FHDI) program supported by NSF CSSI Grant (1931380).
- Inherits the strengths of the serial FHDI R package, thereby requiring no distributional assumptions of data. Also, it inherits parallel computing power of P-FHDI.
- Suitable for curing ultra data (i.e., concurrently big-n and big-p - large instances up to millions with higher dimensions up to 10,000 variables) with complex, irregular missing patterns.
- Examples include continuous and hybrid (categorical & continuous) incomplete data sets.
- Full c++, MPI codes along with sample data are available at
https://iastate.box.com/s/tcl8u50mlk0hh270z2cgscdr3z17azfu
- Ultra data (concurrently big-n and big-p) set is available at
- Science Gateways 2021 Presentation Slide
https://iastate.box.com/s/ssrmherf966t9rqcd1az5fnrb4w1a8q4
Question or bug report: icho@iastate.edu