UP-FHDI: A Software for Big Incomplete Data Curing
Fractional hot-deck imputation (FHDI) is a general-purpose, assumption-free imputation method for handling multivariate missing data by filling each missing item with multiple observed values without resorting to artificially created values. By leveraging FHDI theory and parallel computing techniques, the ultra data-oriented parallel fractional hot-deck imputation (UP-FHDI) was proposed, capable of curing a wide spectrum of big missing data. However, strict pre-process specifications and intricate deployments are major obstacles preventing UP-FHDI from real-world applications. To maximize users’ benefits, we develop a graphical user interface (GUI) to significantly ease the use of UP-FHDI. This paper elaborates on the design process and novel features for improving user experience. Experiments affirm that the proposed software can tackle various real-world incomplete datasets and seamlessly guide users for easy and quick deployment. UP-FHDI will benefit a broad audience in science and engineering without a strong background in coding and imputation.
Committee: Qi Li (major professor) and In-Ho Cho