#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Fri Apr 13 15:01:37 2018
@author: bitzer
"""
import numpy as np
import matplotlib.pyplot as plt
import helpers
from rotated_directions import rotated_directions as dirmodel
#%% create model based on experimental data
data = helpers.load_subject(19, exclude_to=False, censor_late=True)
N = data.shape[0]
choices = [1, -1]
ndtdist = 'uniform'
model = dirmodel(data.tarDir / 180. * np.pi,
helpers.dt,
data.tarDir.unique() / 180. * np.pi,
data.critDir / 180. * np.pi,
maxrt=helpers.maxrt + helpers.dt,
toresponse=helpers.toresponse,
choices=choices,
ndtdist=ndtdist)
model.bound = 0.7
model.intstd = 0.18
exclude_to = store['scalar_opt'].exclude_to
subjects = samples.index.get_level_values('subject').unique()
#%%
diff_data = pd.DataFrame([], columns=['accuracy', 'median RT'], index=subjects)
diff_post = pd.DataFrame([],
columns=['accuracy', 'median RT'],
index=pd.MultiIndex.from_product(
[subjects, np.arange(S)],
names=['subject', 'sample']))
for sub in subjects:
print('\rprocessing subject %2d ...' % sub, end='')
data = helpers.load_subject(sub,
exclude_to=exclude_to,
censor_late=censor_late)
stim = data.response.copy().astype(int)
err = data.error.astype(bool)
stim[err] = -stim[err]
with pd.HDFStore(result, 'r') as store:
ppc_data = store.select('ppc_data', 'subject=sub').loc[sub]
ppc_data.sort_index(inplace=True)
N = ppc_data.loc[0].shape[0]
resh = lambda data: data.values.reshape(N, S, order='F')
diff_post.loc[sub,
'accuracy'], diff_post.loc[sub,
eegdir=helpers.eegdir if use_liks else None)
#subjects = [18]
options = dict(use_liks=use_liks,
censor_late=True,
exclude_to=False,
stats='hist')
resdir = os.path.join(helpers.resultsdir, 'snl', 'rotated_directions',
pd.datetime.now().strftime('%Y%m%d%H%M'))
os.mkdir(resdir)
for sub in subjects:
if options['stats'] == 'id':
data = helpers.load_subject(sub,
exclude_to=options['exclude_to'],
censor_late=options['censor_late'])
conditions = data.easy
else:
conditions = None
sim, stat, data = snlsim.create_simulator(sub,
pars,
ndtdist=ndtdist,
fix=fix,
**options)
# p = pars.sample(10)
# stat.calc(sim.sim(p))
obs_xs = stat.calc(data[['response', 'RT']])
@author: bitzer
"""
import numpy as np
#import pandas as pd
import helpers
import rotated_directions as rtmodels
import pyEPABC
from pyEPABC.parameters import exponential, gaussprob
import matplotlib.pyplot as plt
#%% load data
sub = 18
data = helpers.load_subject(sub)
#%% create model
dt = 0.05
# first of these indicates clockwise rotation, second anti-clockwise
choices = [1, -1]
model = rtmodels.rotated_directions(data.tarDir / 180 * np.pi,
dt,
data.tarDir.unique() / 180 * np.pi,
data.critDir / 180 * np.pi,
maxrt=data.dropna().RT.max() + dt,
toresponse=helpers.toresponse,
choices=choices)
def create_simulator(sub,
pars,
use_liks=False,
stats='hist',
exclude_to=False,
censor_late=True,
ndtdist='lognorm',
fix={}):
data = helpers.load_subject(sub,
exclude_to=exclude_to,
censor_late=censor_late)
# first of these indicates clockwise rotation, second anti-clockwise
choices = [1, -1]
# identify the model based on inferred parameters
modelstr = identify_model(np.r_[pars.names, fix.keys()])
if modelstr == 'diff':
if use_liks is None:
model = dirdiffmodel(
{
'directions': data.tarDir,
'criteria': data.critDir
},
helpers.dt,
maxrt=helpers.maxrt + helpers.dt,
toresponse=helpers.toresponse,
choices=choices,
ndtdist=ndtdist,
**fix)
else:
pass
else:
if use_liks:
Trials, trind, channels = helpers.load_subject_eeg(sub)
directions = -channels % (2 * np.pi)
data = data.reindex(trind)
model = dirmodel(Trials,
dt=helpers.dt,
directions=directions,
criteria=data.critDir / 180. * np.pi,
maxrt=helpers.maxrt + helpers.dt,
toresponse=helpers.toresponse,
choices=choices,
ndtdist=ndtdist,
trial_dirs=data.tarDir / 180. * np.pi,
**fix)
else:
model = dirmodel(data.tarDir / 180. * np.pi,
helpers.dt,
data.tarDir.unique() / 180. * np.pi,
data.critDir / 180. * np.pi,
maxrt=helpers.maxrt + helpers.dt,
toresponse=helpers.toresponse,
choices=choices,
ndtdist=ndtdist,
**fix)
if stats == 'hist':
stats = Stats_hist(model,
pars,
data['easy'],
rts=data.RT,
exclude_to=exclude_to)
elif stats == 'quant':
stats = Stats_quant(model, pars, data['easy'], exclude_to=exclude_to)
elif stats == 'id':
stats = Stats_id(model, pars, data['easy'])
else:
raise ValueError('Unknown type of summary statistics!')
return Simulator(model, pars), stats, data