def initialize_model(Nmax, dim):
"""
Initialize a HSMM model.
Parameters
------
Nmax : int, optional
Maximum number of states
dim : int
The number of channels
Returns
------
model : pyhsmm model
The initial model
"""
obs_hypparams = {
'mu_0': np.zeros(dim), # mean of gaussians
'sigma_0': np.eye(dim), # std of gaussians
'kappa_0': 0.3,
'nu_0': dim + 5
}
# duration is going to be poisson, so prior is a gamma distribution
# (params alpha beta)
expected_lam = 12 * 30
dur_hypparams = {'alpha_0': 2 * expected_lam, 'beta_0': 2}
obs_distns = [
distributions.Gaussian(**obs_hypparams) for state in range(Nmax)
]
dur_distns = [
distributions.PoissonDuration(**dur_hypparams) for state in range(Nmax)
]
model = pyhsmm.models.WeakLimitHDPHSMM(
alpha=6.,
gamma=6., # priors
init_state_concentration=6., # alpha0 for the initial state
obs_distns=obs_distns,
dur_distns=dur_distns)
return model
# Plot the truth
plt.figure()
truemodel.plot()
plt.gcf().suptitle('True HSMM')
########################
# parallel inference #
########################
# building and running the model looks about the same as it usually does
posteriormodel = pyhsmm.models.HSMM(
alpha=6,
gamma=6,
init_state_concentration=6.,
obs_distns=[distns.Gaussian(**obs_hypparams) for state in xrange(N)],
dur_distns=[
distns.PoissonDuration(**dur_hypparams) for state in xrange(N)
])
for data in datas:
posteriormodel.add_data_parallel(data) # NOTE: add_data_parallel
# and so does resampling!
for itr in progprint_xrange(100):
posteriormodel.resample_model_parallel() # NOTE: resample_model_parallel
plt.figure()
posteriormodel.plot()
plt.gcf().suptitle('Sampled model after 100 iterations')
Nmax = 5
obs_hypparams = {'mu_0':np.zeros(obs_dim),
'sigma_0':np.eye(obs_dim),
'kappa_0':0.3,
'nu_0':obs_dim+10
}
# obs_hypparams = {'h_0':np.zeros(obs_dim),
# 'J_0':np.ones(obs_dim) * 0.001, #sq_0 #changes the hidden state detection (the lower the better) #0.001
# 'alpha_0':np.ones(obs_dim) * 0.1, #(make the number of hidden states worse higher the better)
# 'beta_0':np.ones(obs_dim) * 1}
dur_hypparams = {'alpha_0':2*30*500,
'beta_0':3}
obs_distns = [distributions.Gaussian(**obs_hypparams) for state in range(Nmax)]
dur_distns = [distributions.PoissonDuration(**dur_hypparams) for state in range(Nmax)]
datas = online_df.loc[online_df['indicator']==3, ['p_del','p_dup']].values
posteriormodel = pyhsmm.models.WeakLimitHDPHSMM(
# alpha=6.,gamma=6., # better to sample over these; see concentration-resampling.py 10,1
alpha=1.,gamma=1./4,
init_state_concentration=600., # pretty inconsequential
obs_distns=obs_distns,
dur_distns=dur_distns)
posteriormodel.add_data(datas,trunc=80)
for idx in progprint_xrange(40): # 100->50
posteriormodel.resample_model()#num_procs=1)
exp_state = np.empty(len(online_df), dtype=int)
def generate_samples(self, state, num_samples):
dist_parameters = self.emission_params(state)
return distributions.Gaussian(**dist_parameters).rvs(num_samples)