assert len(choice[trial_start:trial_end]) == len(ses_data)
assert np.array_equal(choice[trial_start:trial_end],ses_data['choices'].to_numpy(),equal_nan=True)==True
ses_data.loc[ses_data['choices']]
if trial_end<0:
filling_length_end = abs(trial_end)
else:
filling_length_end = len(choice)-trial_end
filling_length_start = trial_start
end_filler=np.empty(filling_length_end)
end_filler[:] = np.nan
start_filler=np.empty(filling_length_start)
start_filler[:] = np.nan
assert (np.isnan(np.nanmean(ses_data['Qwater'].to_numpy()))==False) & (np.isnan(np.nanmean(ses_data['Qlaser'].to_numpy()))==False)
assert (np.isnan(np.nanmean(ses_data['QRstay'].to_numpy()))==False) & (np.isnan(np.nanmean(ses_data['QLstay'].to_numpy()))==False)
np.save(alf+'/'+prefix+'_choice_prediction.npy', np.concatenate([start_filler,ses_data['predicted_choice'].to_numpy(),end_filler]))
np.save(alf+'/'+prefix+'_water.npy', np.concatenate([start_filler,ses_data['Qwater'].to_numpy(),end_filler]))
np.save(alf+'/'+prefix+'_laser.npy', np.concatenate([start_filler,ses_data['Qlaser'].to_numpy(),end_filler]))
np.save(alf+'/'+prefix+'_QLstay.npy', np.concatenate([start_filler,ses_data['QLstay'].to_numpy(),end_filler]))
np.save(alf+'/'+prefix+'_QRstay.npy', np.concatenate([start_filler,ses_data['QRstay'].to_numpy(),end_filler]))
ROOT_FOLDER = '/Volumes/witten/Alex/Data/Subjects'
psy=mc.load_data(trial_start=0,trial_end=-150)
standata = mc.make_stan_data(psy)
reinforce_params = pd.read_csv('/Volumes/witten/Alex/Data/ephys_bandit/stan_fits/REINFORCE_laserdecaywin_mixedstay/output/summary.csv')
reinforce_data = pd.DataFrame()
reinforce_data = mc.reinforce_model_alphalaserdecay_mixed_perseveration(standata,saved_params=reinforce_params)
assert len(psy) == len(reinforce_data)
prefix='REINFORCE_mixedstay_alphalaserdecay'
save_q_values(ROOT_FOLDER,psy,reinforce_data,prefix, trial_start=0,trial_end=-150)
axis=0)))
WAIC_laserdecay_mixedperseveration = -2 * (
np.sum(np.log(REINFORCE_laserdecaywinloss_mixedstay.mean(axis=0)))
- np.sum(
np.var(np.log(REINFORCE_laserdecaywinloss_mixedstay), axis=0)))
WAIC_laserdecay_mixedperseveration_noloss = -2 * (np.sum(
np.log(REINFORCE_laserdecaywinloss_mixedstay_noloss.mean(axis=0))
) - np.sum(
np.var(np.log(REINFORCE_laserdecaywinloss_mixedstay_noloss),
axis=0)))
WAIC_standard_laserdecay = -2 * (
np.sum(np.log(standard_laserdecay.mean(axis=0))) -
np.sum(np.var(np.log(standard_laserdecay), axis=0)))
except:
# 1
standata = mc.make_stan_data(mc.load_data(trial_end=-1))
path = '/Volumes/witten/Alex/Data/ephys_bandit/stan_fits/standard_reduced_stay'
model = mc.q_learning_model_reduced_stay
WAIC_reducedstay = lppd_from_chain(path,
model,
standata,
save=True,
save_name='qreducedstay')
# 2
path = '/Volumes/witten/Alex/Data/ephys_bandit/stan_fits/standard_reduced'
model = mc.q_learning_model_reduced
WAIC_reduced = lppd_from_chain(path,
model,
standata,
save=True,
save_name='qreduced')