Extract stanfit from model object in Py following cmdstanpy refactoring (#1353)

* add self.stan_fit in fit methods

* small changes to fix tests

* make changes to remove self.params from backend classes
This commit is contained in:
Ryan Nazareth 2020-03-07 02:11:04 +00:00 committed by GitHub
parent 708ae20c04
commit 0ca50cfb7f
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -19,6 +19,7 @@ class IStanBackend(ABC):
def __init__(self, logger): def __init__(self, logger):
self.model = self.load_model() self.model = self.load_model()
self.logger = logger self.logger = logger
self.stan_fit = None
@staticmethod @staticmethod
@abstractmethod @abstractmethod
@ -74,7 +75,7 @@ class CmdStanPyBackend(IStanBackend):
kwargs['algorithm'] = 'Newton' if stan_data['T'] < 100 else 'LBFGS' kwargs['algorithm'] = 'Newton' if stan_data['T'] < 100 else 'LBFGS'
iterations = int(1e4) iterations = int(1e4)
try: try:
stan_fit = self.model.optimize(data=stan_data, self.stan_fit = self.model.optimize(data=stan_data,
inits=stan_init, inits=stan_init,
iter=iterations, iter=iterations,
**kwargs) **kwargs)
@ -85,14 +86,14 @@ class CmdStanPyBackend(IStanBackend):
'Optimization terminated abnormally. Falling back to Newton.' 'Optimization terminated abnormally. Falling back to Newton.'
) )
kwargs['algorithm'] = 'Newton' kwargs['algorithm'] = 'Newton'
stan_fit = self.model.optimize(data=stan_data, self.stan_fit = self.model.optimize(data=stan_data,
inits=stan_init, inits=stan_init,
iter=iterations, iter=iterations,
**kwargs) **kwargs)
else: else:
raise e raise e
params = self.stan_to_dict_numpy(stan_fit.column_names, stan_fit.optimized_params_np) params = self.stan_to_dict_numpy(self.stan_fit.column_names, self.stan_fit.optimized_params_np)
for par in params: for par in params:
params[par] = params[par].reshape((1, -1)) params[par] = params[par].reshape((1, -1))
return params return params
@ -105,14 +106,14 @@ class CmdStanPyBackend(IStanBackend):
if 'warmup_iters' not in kwargs: if 'warmup_iters' not in kwargs:
kwargs['warmup_iters'] = samples // 2 kwargs['warmup_iters'] = samples // 2
stan_fit = self.model.sample(data=stan_data, self.stan_fit = self.model.sample(data=stan_data,
inits=stan_init, inits=stan_init,
sampling_iters=samples, sampling_iters=samples,
**kwargs) **kwargs)
res = stan_fit.sample res = self.stan_fit.sample
(samples, c, columns) = res.shape (samples, c, columns) = res.shape
res = res.reshape((samples * c, columns)) res = res.reshape((samples * c, columns))
params = self.stan_to_dict_numpy(stan_fit.column_names, res) params = self.stan_to_dict_numpy(self.stan_fit.column_names, res)
for par in params: for par in params:
s = params[par].shape s = params[par].shape
@ -220,10 +221,10 @@ class PyStanBackend(IStanBackend):
iter=samples, iter=samples,
) )
args.update(kwargs) args.update(kwargs)
stan_fit = self.model.sampling(**args) self.stan_fit = self.model.sampling(**args)
out = dict() out = dict()
for par in stan_fit.model_pars: for par in self.stan_fit.model_pars:
out[par] = stan_fit[par] out[par] = self.stan_fit[par]
# Shape vector parameters # Shape vector parameters
if par in ['delta', 'beta'] and len(out[par].shape) < 2: if par in ['delta', 'beta'] and len(out[par].shape) < 2:
out[par] = out[par].reshape((-1, 1)) out[par] = out[par].reshape((-1, 1))
@ -239,17 +240,19 @@ class PyStanBackend(IStanBackend):
) )
args.update(kwargs) args.update(kwargs)
try: try:
params = self.model.optimizing(**args) self.stan_fit = self.model.optimizing(**args)
except RuntimeError: except RuntimeError:
# Fall back on Newton # Fall back on Newton
self.logger.warning( self.logger.warning(
'Optimization terminated abnormally. Falling back to Newton.' 'Optimization terminated abnormally. Falling back to Newton.'
) )
args['algorithm'] = 'Newton' args['algorithm'] = 'Newton'
params = self.model.optimizing(**args) self.stan_fit = self.model.optimizing(**args)
for par in params: params = dict()
params[par] = params[par].reshape((1, -1))
for par in self.stan_fit.keys():
params[par] = self.stan_fit[par].reshape((1, -1))
return params return params