[1]Department of Plant and Environmental Sciences, University of Copenhagen, Denmark;[2]Department of Food Science, University of Copenhagen, Denmark
INDAFAC (INcomplete DAta paraFAC) is a sparse Levenberg-Marquardt algorithm for fitting PARAFAC to three-way arrays with missing data. Supports multiple convergence criteria and initialization methods (random, SVD/NIPALS, DTLD/GRAM). Includes CreaMiss.m for generating synthetic incomplete arrays. The companion PARAFAC3W package provides Levenberg-Marquardt, ALS, SWATLD, and PMF3 algorithms with optional compression.
References:
Tomasi, G.; Bro, R. (2005). PARAFAC missing values. Chemometrics and Intelligent Laboratory Systems, 75(2), 163-180. https://doi.org/10.1016/j.chemolab.2004.07.003
Tomasi, G.; Bro, R. (2006). A comparison of algorithms for fitting the PARAFAC model. Computational Statistics & Data Analysis, 50, 1700-1734. https://doi.org/10.1016/j.csda.2004.11.013