Lapse rates for model fitsΒΆ
When fitting models, especially when doing so with likelihood, it is
useful to have a constant lapse rate in the model to prevent the
likelihood from being negative infinity. PyDDM has two useful
built-in lapse rates for this which are used as mixture models: an
Exponential lapse rate
(according
to a Poisson process, the recommended method) and the Uniform
lapse rate
(which is more common in the
literature). These can be introduced with:
from pyddm import Model
from pyddm.models import OverlayPoissonMixture, OverlayUniformMixture
model1 = Model(overlay=OverlayPoissonMixture(pmixturecoef=.05, rate=1))
model2 = Model(overlay=OverlayUniformMixture(umixturecoef=.05))
If another overlay is to be used, such as
OverlayNonDecision
, then an OverlayChain
object
must be used:
from pyddm import Model
from pyddm.models import OverlayPoissonMixture, OverlayNonDecision, OverlayChain
model = Model(overlay=OverlayChain(overlays=[OverlayNonDecision(nondectime=.2),
OverlayPoissonMixture(pmixturecoef=.05, rate=1)]))