Closed erikbern closed 6 years ago
🤔 still looks like there's a mild bias in the distribution... not sure what's up
This seems like a marginal improvement but I'm still not 100% sure about the nonparametric model... will investigate more. Merging for now
Think I figured out what's up. If z ~ N(m, s)
and m < 0
then E(sigmoid(z)) > sigmoid(m)
so you end up accumulating a bunch of positive biases. The MLE actually doesn't have the same problem.
Will think of some way to fix this, probably entails getting rid of the cumsum
stuff
Found a few issues
numpy.random.normal_multivariate
to complain and generate bogus values. This is despite eigenvalues being all positivez
thatexpit
can't handle. Clipping fixes that problemn=1000
its end up having a fairly substantial upwards bias early in the distribution, whereas this disappears forn=100
. Not quite sure what's up.After these changes, Weibull estimation of synthetic Weibull data lines up fairly well with the nonparametric estimation