def test_reconstruct():
"""Test Tracker objects reconstruction"""
# test with one TrackerUD
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = TrackerUD(ec, 1, 1, 3, 1, 5, np.inf, 3)
while not tr.stopped:
tr.respond(np.random.rand() < tr.x_current)
tracker = reconstruct_tracker(ec.data_fname)[0]
assert_true(tracker.stopped)
tracker.x_current
# test with one TrackerBinom
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = TrackerBinom(ec, .05, .5, 10)
while not tr.stopped:
tr.respond(True)
tracker = reconstruct_tracker(ec.data_fname)[0]
assert_true(tracker.stopped)
tracker.x_current
# tracker not stopped
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = TrackerUD(ec, 1, 1, 3, 1, 5, np.inf, 3)
tr.respond(np.random.rand() < tr.x_current)
assert_true(not tr.stopped)
assert_raises(ValueError, reconstruct_tracker, ec.data_fname)
# test with dealer
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = [TrackerUD(ec, 1, 1, 3, 1, 5, np.inf, 3) for _ in range(3)]
td = TrackerDealer(ec, tr)
for _, x_current in td:
td.respond(np.random.rand() < x_current)
dealer = reconstruct_dealer(ec.data_fname)[0]
assert_true(all(td._x_history == dealer._x_history))
assert_true(all(td._tracker_history == dealer._tracker_history))
assert_true(all(td._response_history == dealer._response_history))
assert_true(td.shape == dealer.shape)
assert_true(td.trackers.shape == dealer.trackers.shape)
# no tracker/dealer in file
with ExperimentController(*std_args, **std_kwargs) as ec:
ec.identify_trial(ec_id='one', ttl_id=[0])
ec.start_stimulus()
ec.write_data_line('misc', 'trial one')
ec.stop()
ec.trial_ok()
ec.write_data_line('misc', 'end')
assert_raises(ValueError, reconstruct_tracker, ec.data_fname)
assert_raises(ValueError, reconstruct_dealer, ec.data_fname)
for ss, level in td:
# Get information of which trial type is next and what the level is at
# that time from TrackerDealer
td.respond(rng_human.rand() < sigmoid(
level - true_thresh[sum(ss)], lower=chance, slope=slope))
###############################################################################
# Reconstructing the TrackerDealer Object
# ---------------------------------------
# The TrackerDealer object has many built in analysis functions that are can
# only be access through the object itself (not the log files alone). By using
# :func:`expyfun.io.reconstruct_dealer`, the object can be recreated such that
# the analysis functions are accessible. Note that the function always returns
# a list of objects. Similar reconstructions of single trackers can be done
# with :func:`expyfun.io.reconstruct_tracker`.
td_tab = reconstruct_dealer(ec.data_fname)[0]
##############################################################################
# Plotting the Results
# ---------------------------
axes = plt.subplots(2, 1)[1]
for i in [0, 1]:
fig, ax, lines = td_tab.trackers.ravel()[i].plot(ax=axes[i], n_skip=4)
ax.legend(loc='best')
ax.set_title('Adaptive track of model human trial type {} (true threshold '
'is {})'.format(i + 1, true_thresh[i]))
fig.tight_layout()
'027', '029', '030', '031', '032', '034', '036', '037',
'038', '039', '042', '043']
pvals = []
d_prime = []
per_cor = []
diff = []
diff_per = []
responses = []
biases = []
ind = 1
for subject, fname in zip(subjects, fnames):
print(subject)
# %% read, reformat data
td = reconstruct_dealer(path + fname + '.tab')[0]
conditions = [1.25, 2.5, 5, 10, 20]
x = [np.concatenate([-td.trackers.flat[2*ii].x, td.trackers.flat[2*ii +1].x]) for ii in range(len(conditions) * 2)]
responses.append([np.concatenate([np.expand_dims(np.asarray(td.trackers.flat[2* ii].responses, dtype=bool), 1),
np.expand_dims(np.asarray(td.trackers.flat[2 * ii + 1].responses, dtype=bool), 1)], 1) for ii in range(len(conditions * 2))])
per_cor.append(get_per_correct(responses[-1]))
percent = [per_cor[-1][:5], per_cor[-1][5:]]
# %% do fits
data = np.array(percent) / 100.
lower = 0.5
upper = 1
conditions = 20 * np.log10(conditions)
for ss, level in td:
# Get information of which trial type is next and what the level is at
# that time from TrackerDealer
td.respond(rng_human.rand() < sigmoid(level - true_thresh[sum(ss)],
lower=chance, slope=slope))
###############################################################################
# Reconstructing the TrackerDealer Object
# ---------------------------------------
# The TrackerDealer object has many built in analysis functions that are can
# only be access through the object itself (not the log files alone). By using
# :func:`expyfun.io.reconstruct_dealer`, the object can be recreated such that
# the analysis functions are accessible. Note that the function always returns
# a list of objects. Similar reconstructions of single trackers can be done
# with :func:`expyfun.io.reconstruct_tracker`.
td_tab = reconstruct_dealer(ec.data_fname)[0]
##############################################################################
# Plotting the Results
# ---------------------------
axes = plt.subplots(2, 1)[1]
for i in [0, 1]:
fig, ax, lines = td_tab.trackers.ravel()[i].plot(ax=axes[i], n_skip=4)
ax.legend(loc='best')
ax.set_title('Adaptive track of model human trial type {} (true threshold '
'is {})'.format(i + 1, true_thresh[i]))
fig.tight_layout()
