This is an alternative to predictValues that assumes binary classification and assumes that the model won't be asked to use an out-of-bounds feature, both of which cut down on overhead.
This is an alternative to predictValues that assumes binary classification and assumes that the model won't be asked to use an out-of-bounds feature, both of which cut down on overhead. It will give out-of-bounds exceptions if not used appropriately, and will not work at all for polytomous models.
This is how we'd prefer to write predictValuesBinary, but unfortunately, the while loop is faster than the fold (by about a factor of two, it seems).
This is how we'd prefer to write predictValuesBinary, but unfortunately, the while loop is faster than the fold (by about a factor of two, it seems). :(
A simple wrapper to Liblinear.