nems.segmentors module

nems.segmentors.use_all_data(data)[source]

Returns a segmentor function (see fitter.py and docs/fitters.md). A pass-thru function. Use this when you want to fit on all of your estimation set data, all of the time.

nems.segmentors.random_jackknife_maker(nsplits=10, rebuild_every=50, invert=False, excise=False)[source]

Returns a segmentor function (see fitter.py and docs/fitters.md).

This segmentor takes a random jackknife of the (estimation set) data to fit on for rebuild_every evaluations of the cost function. The optional argument nsplits determines how many pieces to split the data into before picking one of the pieces at random. The optional argument rebuild_every decides how many calls to mylambda before the subset is rebuilt.