search

cameronraysmith / eeda

0 stars 0 forks source link

generalize to multiple observations #1

Closed cameronraysmith closed 11 years ago

cameronraysmith commented 11 years ago

Currently gca.py only works for a single observation at a time. Independent parameter estimates for the modified Gompertz growth function could be obtained by looping over the observation vectors. Try using the suggestion here to generalize to multiple observations.

cameronraysmith commented 11 years ago

One can vectorize fitting independent runs using multivariate distributions as suggested here.

9ea26fe takes into account multiple data points (i.e. replicates) at each time point by concatenating the replicates and replicating the corresponding time vector. This resulted in some autocorrelation in the mcmc for the lag time and linear slope parameters. This can be improved to some degree by increasing the burntime from 5000 to 95000 iterations.