Scripts to munge certain machine learning sparse data formats, specifically the common pattern of a sparse feature matrix plus a column of labels/predictions.
Formats are
sl
-- the SVMLight/LibSVM format, loosely as described at https://svmlight.joachims.org/coord
-- 1-indexed i,j,val coordinate-format triples. (matlab-compatible)mm
-- MatrixMarket header file, as required by R's Matrix package. https://math.nist.gov/MatrixMarket/formats.htmlmegam
-- one of the variants described at https://www.cs.utah.edu/~hal/megam/dense
-- delimited