def run(self):
subject = self.pipeline.subject
task = self.pipeline.task
events = self.get_passed_object(task + '_events')
math_events = self.get_passed_object(task + '_math_events')
intr_events = self.get_passed_object(task + '_intr_events')
rec_events = self.get_passed_object(task + '_rec_events')
all_events = self.get_passed_object(task + '_all_events')
monopolar_channels = self.get_passed_object('monopolar_channels')
bipolar_pairs = self.get_passed_object('bipolar_pairs')
loc_info = self.get_passed_object('loc_info')
ttest = self.get_passed_object('ttest')
xval_output = self.get_passed_object('xval_output')
perm_test_pvalue = self.get_passed_object('pvalue')
sessions = np.unique(events.session)
self.pass_object('NUMBER_OF_SESSIONS', len(sessions))
self.pass_object('NUMBER_OF_ELECTRODES', len(monopolar_channels))
session_data = []
session_summary_array = []
session_ttest_data = []
positions = np.unique(events.serialpos)
first_recall_counter = np.zeros(positions.size, dtype=int)
total_list_counter = 0
irt_within_cat = []
irt_between_cat = []
for session in sessions:
session_summary = SessionSummary()
session_events = events[events.session == session]
n_sess_events = len(session_events)
session_rec_events = rec_events[rec_events.session == session]
session_all_events = all_events[all_events.session == session]
timestamps = session_all_events.mstime
first_time_stamp = np.min(timestamps)
last_time_stamp = np.max(timestamps)
session_length = '%.2f' % (
(last_time_stamp - first_time_stamp) / 60000.0)
session_date = time.strftime(
'%d-%b-%Y', time.localtime(last_time_stamp / 1000))
session_data.append([session, session_date, session_length])
session_name = 'Sess%02d' % session
print 'Session =', session_name
session_summary.number = session
session_summary.name = session_name
session_summary.length = session_length
session_summary.date = session_date
session_summary.n_words = len(session_events)
session_summary.n_correct_words = np.sum(session_events.recalled)
session_summary.pc_correct_words = 100 * session_summary.n_correct_words / float(
session_summary.n_words)
positions = np.unique(session_events.serialpos)
prob_recall = np.empty_like(positions, dtype=float)
for i, pos in enumerate(positions):
pos_events = session_events[session_events.serialpos == pos]
prob_recall[i] = np.sum(pos_events.recalled) / float(
len(pos_events))
session_summary.prob_recall = prob_recall
lists = np.unique(session_events.list)
n_lists = len(lists)
prob_first_recall = np.zeros(len(positions), dtype=float)
session_irt_within_cat = []
session_irt_between_cat = []
for lst in lists:
list_rec_events = session_rec_events[
(session_rec_events.list == lst)
& (session_rec_events.intrusion == 0)]
if list_rec_events.size > 0:
list_events = session_events[session_events.list == lst]
tmp = np.where(
list_events.itemno == list_rec_events[0].itemno)[0]
if tmp.size > 0:
first_recall_idx = tmp[0]
prob_first_recall[first_recall_idx] += 1
first_recall_counter[first_recall_idx] += 1
if task == 'RAM_CatFR1':
# list_rec_events = session_rec_events[session_rec_events.list == lst]
for i in xrange(1, len(list_rec_events)):
cur_ev = list_rec_events[i]
prev_ev = list_rec_events[i - 1]
# if (cur_ev.intrusion == 0) and (prev_ev.intrusion == 0):
dt = cur_ev.mstime - prev_ev.mstime
if cur_ev.category == prev_ev.category:
session_irt_within_cat.append(dt)
else:
session_irt_between_cat.append(dt)
prob_first_recall /= float(n_lists)
total_list_counter += n_lists
session_summary.irt_within_cat = sum(session_irt_within_cat) / len(
session_irt_within_cat) if session_irt_within_cat else 0.0
session_summary.irt_between_cat = sum(
session_irt_between_cat) / len(
session_irt_between_cat
) if session_irt_between_cat else 0.0
irt_within_cat += session_irt_within_cat
irt_between_cat += session_irt_between_cat
session_summary.prob_first_recall = prob_first_recall
if math_events is not None:
session_math_events = math_events[math_events.session ==
session]
session_summary.n_math = len(session_math_events)
session_summary.n_correct_math = np.sum(
session_math_events.iscorrect)
session_summary.pc_correct_math = 100 * session_summary.n_correct_math / float(
session_summary.n_math)
session_summary.math_per_list = session_summary.n_math / float(
n_lists)
session_intr_events = intr_events[intr_events.session == session]
session_summary.n_pli = np.sum(session_intr_events.intrusion > 0)
session_summary.pc_pli = 100 * session_summary.n_pli / float(
n_sess_events)
session_summary.n_eli = np.sum(session_intr_events.intrusion == -1)
session_summary.pc_eli = 100 * session_summary.n_eli / float(
n_sess_events)
session_xval_output = xval_output[session]
session_summary.auc = '%.2f' % (100 * session_xval_output.auc)
session_summary.fpr = session_xval_output.fpr
session_summary.tpr = session_xval_output.tpr
session_summary.pc_diff_from_mean = (
session_xval_output.low_pc_diff_from_mean,
session_xval_output.mid_pc_diff_from_mean,
session_xval_output.high_pc_diff_from_mean)
session_summary_array.append(session_summary)
# ttest_data = [list(a) for a in zip(bipolar_pairs.eType, bipolar_pairs.tagName, ttest[session][1], ttest[session][0])]
session_ttest = ttest[session]
if isinstance(session_ttest, tuple):
if ('Das Volumetric Atlas Location' in loc_info) or (
'Freesurfer Desikan Killiany Surface Atlas Location'
in loc_info):
session_ttest_data.append(
[[None, None, None, None, np.nan, np.nan]])
else:
session_ttest_data.append([[None, None, np.nan, np.nan]])
else:
ttest_data = make_ttest_table(bipolar_pairs, loc_info,
session_ttest)
ttest_data.sort(key=itemgetter(-2))
ttest_data = format_ttest_table(ttest_data)
session_ttest_data.append(ttest_data)
self.pass_object('SESSION_DATA', session_data)
self.pass_object('session_summary_array', session_summary_array)
self.pass_object('session_ttest_data', session_ttest_data)
cumulative_summary = SessionSummary()
cumulative_summary.n_words = len(events)
cumulative_summary.n_correct_words = np.sum(events.recalled)
cumulative_summary.pc_correct_words = 100 * cumulative_summary.n_correct_words / float(
cumulative_summary.n_words)
cumulative_summary.irt_within_cat = sum(irt_within_cat) / len(
irt_within_cat) if irt_within_cat else 0.0
cumulative_summary.irt_between_cat = sum(irt_between_cat) / len(
irt_between_cat) if irt_between_cat else 0.0
positions = np.unique(events.serialpos)
prob_recall = np.empty_like(positions, dtype=float)
for i, pos in enumerate(positions):
pos_events = events[events.serialpos == pos]
prob_recall[i] = np.sum(pos_events.recalled) / float(
len(pos_events))
cumulative_summary.prob_recall = prob_recall
prob_first_recall = first_recall_counter / float(total_list_counter)
cumulative_summary.prob_first_recall = prob_first_recall
if math_events is not None:
cumulative_summary.n_math = len(math_events)
cumulative_summary.n_correct_math = np.sum(math_events.iscorrect)
cumulative_summary.pc_correct_math = 100 * cumulative_summary.n_correct_math / float(
cumulative_summary.n_math)
cumulative_summary.math_per_list = cumulative_summary.n_math / float(
total_list_counter)
cumulative_summary.n_pli = np.sum(intr_events.intrusion > 0)
cumulative_summary.pc_pli = 100 * cumulative_summary.n_pli / float(
len(events))
cumulative_summary.n_eli = np.sum(intr_events.intrusion == -1)
cumulative_summary.pc_eli = 100 * cumulative_summary.n_eli / float(
len(events))
cumulative_xval_output = xval_output[-1]
cumulative_summary.auc = '%.2f' % (100 * cumulative_xval_output.auc)
cumulative_summary.fpr = cumulative_xval_output.fpr
cumulative_summary.tpr = cumulative_xval_output.tpr
cumulative_summary.pc_diff_from_mean = (
cumulative_xval_output.low_pc_diff_from_mean,
cumulative_xval_output.mid_pc_diff_from_mean,
cumulative_xval_output.high_pc_diff_from_mean)
cumulative_summary.perm_AUCs = self.get_passed_object('perm_AUCs')
cumulative_summary.perm_test_pvalue = (
'= %.3f' %
perm_test_pvalue) if perm_test_pvalue >= 0.001 else '\leq 0.001'
cumulative_summary.jstat_thresh = '%.3f' % cumulative_xval_output.jstat_thresh
cumulative_summary.jstat_percentile = '%.2f' % (
100.0 * cumulative_xval_output.jstat_quantile)
self.pass_object('cumulative_summary', cumulative_summary)
# cumulative_ttest_data = [list(a) for a in zip(bipolar_pairs.eType, bipolar_pairs.tagName, ttest[-1][1], ttest[-1][0])]
cumulative_ttest_data = make_ttest_table(bipolar_pairs, loc_info,
ttest[-1])
cumulative_ttest_data.sort(key=itemgetter(-2))
cumulative_ttest_data = format_ttest_table(cumulative_ttest_data)
self.pass_object('cumulative_ttest_data', cumulative_ttest_data)
ttable_format, ttable_header = make_ttest_table_header(loc_info)
self.pass_object('ttable_format', ttable_format)
self.pass_object('ttable_header', ttable_header)
def run(self):
subject = self.pipeline.subject
experiment = self.pipeline.experiment
monopolar_channels = self.get_passed_object('monopolar_channels')
xval_output = self.get_passed_object('xval_output')
ps_table = self.get_passed_object('ps_table')
sessions = sorted(ps_table.session.unique())
self.pass_object('NUMBER_OF_SESSIONS', len(sessions))
self.pass_object('NUMBER_OF_ELECTRODES', len(monopolar_channels))
thresh = xval_output[-1].jstat_thresh
self.pass_object('AUC', xval_output[-1].auc)
param1_name = param2_name = None
param1_unit = param2_unit = None
const_param_name = const_unit = None
if experiment == 'PS1':
param1_name = 'Pulse_Frequency'
param2_name = 'Duration'
param1_unit = 'Hz'
param2_unit = 'ms'
const_param_name = 'Amplitude'
const_unit = 'mA'
elif experiment == 'PS2':
param1_name = 'Pulse_Frequency'
param2_name = 'Amplitude'
param1_unit = 'Hz'
param2_unit = 'mA'
const_param_name = 'Duration'
const_unit = 'ms'
elif experiment == 'PS3':
param1_name = 'Burst_Frequency'
param2_name = 'Pulse_Frequency'
param1_unit = 'Hz'
param2_unit = 'Hz'
const_param_name = 'Duration'
const_unit = 'ms'
self.pass_object('param1_name', param1_name.replace('_', ' '))
self.pass_object('param1_unit', param1_unit)
self.pass_object('param2_name', param2_name.replace('_', ' '))
self.pass_object('param2_unit', param2_unit)
self.pass_object('const_param_name', const_param_name)
self.pass_object('const_unit', const_unit)
session_data = []
session_summary_array = []
anova_param1_sv = dict()
anova_param2_sv = dict()
anova_param12_sv = dict()
anova_significance = dict()
for session in sessions:
ps_session_table = ps_table[ps_table.session==session]
session_summary = SessionSummary()
session_summary.sess_num = session
first_time_stamp = ps_session_table.mstime.min()
last_time_stamp = ps_session_table.mstime.max()
session_length = '%.2f' % ((last_time_stamp - first_time_stamp) / 60000.0)
session_date = time.strftime('%d-%b-%Y', time.localtime(last_time_stamp/1000))
session_data.append([session, session_date, session_length])
session_name = 'Sess%02d' % session
stim_anode_tag = ps_session_table.stimAnodeTag.values[0].upper()
stim_cathode_tag = ps_session_table.stimCathodeTag.values[0].upper()
stim_tag = stim_anode_tag + '-' + stim_cathode_tag
sess_loc_tag = ps_session_table.Region.values[0]
roi = '{\em locTag not found}' if sess_loc_tag is None else sess_loc_tag
isi_min = ps_session_table.isi.min()
isi_max = ps_session_table.isi.max()
isi_mid = (isi_max+isi_min) / 2.0
isi_halfrange = isi_max - isi_mid
print 'Session =', session_name, ' StimTag =', stim_tag, ' ISI =', isi_mid, '+/-', isi_halfrange
session_summary.name = session_name
session_summary.length = session_length
session_summary.date = session_date
session_summary.stimtag = stim_tag
session_summary.region_of_interest = roi
session_summary.isi_mid = isi_mid
session_summary.isi_half_range = isi_halfrange
session_summary.const_param_value = ps_session_table[const_param_name].unique().max()
ps_session_low_table = pd.DataFrame(ps_session_table[ps_session_table['prob_pre']<thresh])
session_summary.plots = delta_plot_data(ps_session_low_table, param1_name, param2_name, param2_unit)
if sess_loc_tag is not None and not (sess_loc_tag in anova_param1_sv):
anova_param1_sv[sess_loc_tag] = []
anova_param2_sv[sess_loc_tag] = []
anova_param12_sv[sess_loc_tag] = []
anova = anova_test(ps_session_low_table, param1_name, param2_name)
if anova is not None:
session_summary.anova_fvalues = anova[0]
session_summary.anova_pvalues = anova[1]
if anova[1][0] < 0.05: # first param significant
if stim_tag in anova_significance:
anova_significance[stim_tag][0] = True
else:
anova_significance[stim_tag] = np.array([True, False, False], dtype=np.bool)
param1_ttest_table = ttest_one_param(ps_session_low_table, param1_name)
if len(param1_ttest_table) > 0:
if sess_loc_tag is not None:
anova_param1_sv[sess_loc_tag].append(param1_ttest_table)
session_summary.param1_ttest_table = format_ttest_table(param1_ttest_table)
if anova[1][1] < 0.05: # second param significant
if stim_tag in anova_significance:
anova_significance[stim_tag][1] = True
else:
anova_significance[stim_tag] = np.array([False, True, False], dtype=np.bool)
param2_ttest_table = ttest_one_param(ps_session_low_table, param2_name)
if len(param2_ttest_table) > 0:
if sess_loc_tag is not None:
anova_param2_sv[sess_loc_tag].append(param2_ttest_table)
session_summary.param2_ttest_table = format_ttest_table(param2_ttest_table)
if anova[1][2] < 0.05: # interaction is significant
if stim_tag in anova_significance:
anova_significance[stim_tag][2] = True
else:
anova_significance[stim_tag] = np.array([False, False, True], dtype=np.bool)
param12_ttest_table = ttest_interaction(ps_session_low_table, param1_name, param2_name)
if len(param12_ttest_table) > 0:
if sess_loc_tag is not None:
anova_param12_sv[sess_loc_tag].append(param12_ttest_table)
session_summary.param12_ttest_table = format_ttest_table(param12_ttest_table)
session_summary_array.append(session_summary)
self.pass_object('SESSION_DATA', session_data)
self.pass_object('session_summary_array', session_summary_array)
joblib.dump(anova_param1_sv, self.get_path_to_resource_in_workspace(subject + '-' + experiment + '-anova_%s_sv.pkl'%param1_name))
joblib.dump(anova_param2_sv, self.get_path_to_resource_in_workspace(subject + '-' + experiment + '-anova_%s_sv.pkl'%param2_name))
joblib.dump(anova_param12_sv, self.get_path_to_resource_in_workspace(subject + '-' + experiment + '-anova_%s-%s_sv.pkl'%(param1_name,param2_name)))
if len(anova_significance) > 0:
joblib.dump(anova_significance, self.get_path_to_resource_in_workspace(subject + '-' + experiment + '-anova_significance.pkl'))
isi_min = ps_table.isi.min()
isi_max = ps_table.isi.max()
isi_mid = (isi_max+isi_min) / 2.0
isi_halfrange = isi_max - isi_mid
print 'ISI =', isi_mid, '+/-', isi_halfrange
self.pass_object('CUMULATIVE_ISI_MID', isi_mid)
self.pass_object('CUMULATIVE_ISI_HALF_RANGE', isi_halfrange)
ps_low_table = ps_table[ps_table['prob_pre']<thresh]
cumulative_plots = delta_plot_data(ps_low_table, param1_name, param2_name, param2_unit)
self.pass_object('cumulative_plots', cumulative_plots)
cumulative_anova_fvalues = cumulative_anova_pvalues = None
cumulative_param1_ttest_table = cumulative_param2_ttest_table = cumulative_param12_ttest_table = None
anova = anova_test(ps_low_table, param1_name, param2_name)
if anova is not None:
cumulative_anova_fvalues = anova[0]
cumulative_anova_pvalues = anova[1]
if anova[1][0] < 0.05:
param1_ttest_table = ttest_one_param(ps_low_table, param1_name)
if len(param1_ttest_table) > 0:
cumulative_param1_ttest_table = format_ttest_table(param1_ttest_table)
if anova[1][1] < 0.05:
param2_ttest_table = ttest_one_param(ps_low_table, param2_name)
if len(param2_ttest_table) > 0:
cumulative_param2_ttest_table = format_ttest_table(param2_ttest_table)
if anova[1][2] < 0.05:
param12_ttest_table = ttest_interaction(ps_low_table, param1_name, param2_name)
if len(param12_ttest_table) > 0:
cumulative_param12_ttest_table = format_ttest_table(param12_ttest_table)
self.pass_object('CUMULATIVE_ANOVA_FVALUES', cumulative_anova_fvalues)
self.pass_object('CUMULATIVE_ANOVA_PVALUES', cumulative_anova_pvalues)
self.pass_object('CUMULATIVE_PARAM1_TTEST_TABLE', cumulative_param1_ttest_table)
self.pass_object('CUMULATIVE_PARAM2_TTEST_TABLE', cumulative_param2_ttest_table)
self.pass_object('CUMULATIVE_PARAM12_TTEST_TABLE', cumulative_param12_ttest_table)
def run(self):
subject = self.pipeline.subject
task = self.pipeline.task
events = self.get_passed_object(task + '_events')
math_events = self.get_passed_object(task + '_math_events')
intr_events = self.get_passed_object(task + '_intr_events')
rec_events = self.get_passed_object(task + '_rec_events')
all_events = self.get_passed_object(task + '_all_events')
channels = self.get_passed_object('monopolar_channels')
tal_info = self.get_passed_object('bipolar_pairs')
sessions = np.unique(events.session)
self.pass_object('NUMBER_OF_SESSIONS', len(sessions))
self.pass_object('NUMBER_OF_ELECTRODES', len(channels))
session_data = []
session_summary_array = []
positions = np.unique(events.serialpos)
first_recall_counter = np.zeros(positions.size, dtype=int)
total_list_counter = 0
irt_within_cat = []
irt_between_cat = []
cumulative_n_items_from_stim = 0
cumulative_n_recalls_from_stim = 0
cumulative_n_intr_from_stim = 0
cumulative_n_items_from_nonstim = 0
cumulative_n_recalls_from_nonstim = 0
cumulative_n_intr_from_nonstim = 0
for session in sessions:
session_summary = SessionSummary()
session_events = events[events.session == session]
n_sess_events = len(session_events)
session_rec_events = rec_events[rec_events.session == session]
session_all_events = all_events[all_events.session == session]
timestamps = session_all_events.mstime
first_time_stamp = np.min(timestamps)
last_time_stamp = np.max(timestamps)
session_length = '%.2f' % ((last_time_stamp - first_time_stamp) / 60000.0)
session_date = time.strftime('%d-%b-%Y', time.localtime(last_time_stamp/1000))
session_data.append([session, session_date, session_length])
session_name = 'Sess%02d' % session
print 'Session =', session_name
session_summary.number = session
session_summary.name = session_name
session_summary.length = session_length
session_summary.date = session_date
session_summary.n_words = len(session_events)
session_summary.n_correct_words = np.sum(session_events.recalled)
session_summary.pc_correct_words = 100*session_summary.n_correct_words / float(session_summary.n_words)
positions = np.unique(session_events.serialpos)
prob_recall = np.empty_like(positions, dtype=float)
for i,pos in enumerate(positions):
pos_events = session_events[session_events.serialpos == pos]
prob_recall[i] = np.sum(pos_events.recalled) / float(len(pos_events))
session_summary.prob_recall = prob_recall
lists = np.unique(session_events.list)
n_lists = len(lists)
prob_first_recall = np.zeros(len(positions), dtype=float)
session_irt_within_cat = []
session_irt_between_cat = []
session_summary.n_recalls_per_list = np.zeros(n_lists, dtype=np.int)
session_summary.n_intr_per_list = np.zeros(n_lists, dtype=np.int)
session_summary.n_stims_per_list = np.zeros(n_lists, dtype=np.int)
session_summary.is_stim_list = np.zeros(n_lists, dtype=np.bool)
items_per_list = np.zeros(n_lists, dtype=np.int)
for lst in lists:
list_events = session_all_events[session_all_events.list == lst]
list_rec_events = session_rec_events[(session_rec_events.list == lst) & (session_rec_events.intrusion == 0)]
list_intr_events = session_rec_events[(session_rec_events.list == lst) & (session_rec_events.intrusion >= 4)]
list_word_events = list_events[list_events.type=='WORD']
#session_summary.n_recalls_per_list[lst-1] = len(list_rec_events)
#session_summary.n_intr_per_list[lst-1] = len(list_intr_events)
session_summary.n_recalls_per_list[lst-1] = np.sum(list_word_events.recalled)
session_summary.n_stims_per_list[lst-1] = np.sum(list_events.type=='STIM')
session_summary.is_stim_list[lst-1] = session_events[session_events.list == lst][0].stimList
for ie in list_intr_events:
session_summary.n_intr_per_list[ie.intrusion-1] += 1
items_per_list[lst-1] = np.sum(list_events.type=='WORD')
if list_rec_events.size > 0:
list_events = session_events[session_events.list == lst]
tmp = np.where(list_events.itemno == list_rec_events[0].itemno)[0]
if tmp.size > 0:
first_recall_idx = tmp[0]
prob_first_recall[first_recall_idx] += 1
first_recall_counter[first_recall_idx] += 1
if task == 'RAM_CatFR3':
# list_rec_events = session_rec_events[session_rec_events.list == lst]
for i in xrange(1,len(list_rec_events)):
cur_ev = list_rec_events[i]
prev_ev = list_rec_events[i-1]
# if (cur_ev.intrusion == 0) and (prev_ev.intrusion == 0):
dt = cur_ev.mstime - prev_ev.mstime
if cur_ev.category == prev_ev.category:
session_irt_within_cat.append(dt)
else:
session_irt_between_cat.append(dt)
prob_first_recall /= float(n_lists)
total_list_counter += n_lists
n_items_from_stim = np.sum(items_per_list[session_summary.is_stim_list])
n_recalls_from_stim = np.sum(session_summary.n_recalls_per_list[session_summary.is_stim_list])
n_intr_from_stim = np.sum(session_summary.n_intr_per_list[session_summary.is_stim_list])
cumulative_n_items_from_stim += n_items_from_stim
cumulative_n_recalls_from_stim += n_recalls_from_stim
cumulative_n_intr_from_stim += n_intr_from_stim
nonstim_list_mask = ~session_summary.is_stim_list
nonstim_list_mask[0:3] = False
n_items_from_nonstim = np.sum(items_per_list[nonstim_list_mask])
n_recalls_from_nonstim = np.sum(session_summary.n_recalls_per_list[nonstim_list_mask])
n_intr_from_nonstim = np.sum(session_summary.n_intr_per_list[nonstim_list_mask])
cumulative_n_items_from_nonstim += n_items_from_nonstim
cumulative_n_recalls_from_nonstim += n_recalls_from_nonstim
cumulative_n_intr_from_nonstim += n_intr_from_nonstim
session_summary.n_correct_stim = n_recalls_from_stim
session_summary.n_total_stim = n_items_from_stim
session_summary.pc_from_stim = 100 * n_recalls_from_stim / float(n_items_from_stim)
session_summary.n_correct_nonstim = n_recalls_from_nonstim
session_summary.n_total_nonstim = n_items_from_nonstim
session_summary.pc_from_nonstim = 100 * n_recalls_from_nonstim / float(n_items_from_nonstim)
session_summary.n_stim_intr = n_intr_from_stim
session_summary.pc_from_stim_intr = 100 * n_intr_from_stim / float(n_items_from_stim)
session_summary.n_nonstim_intr = n_intr_from_nonstim
session_summary.pc_from_nonstim_intr = 100 * n_intr_from_nonstim / float(n_items_from_nonstim)
session_summary.chisqr,session_summary.pvalue,_ = proportions_chisquare([n_recalls_from_stim, n_recalls_from_nonstim], [n_items_from_stim, n_items_from_nonstim])
session_summary.chisqr_intr,session_summary.pvalue_intr,_ = proportions_chisquare([n_intr_from_stim, n_intr_from_nonstim], [n_items_from_stim, n_items_from_nonstim])
session_summary.irt_within_cat = sum(session_irt_within_cat) / len(session_irt_within_cat) if session_irt_within_cat else 0.0
session_summary.irt_between_cat = sum(session_irt_between_cat) / len(session_irt_between_cat) if session_irt_between_cat else 0.0
irt_within_cat += session_irt_within_cat
irt_between_cat += session_irt_between_cat
session_summary.prob_first_recall = prob_first_recall
if math_events is not None:
session_math_events = math_events[math_events.session == session]
session_summary.n_math = len(session_math_events)
session_summary.n_correct_math = np.sum(session_math_events.iscorrect)
session_summary.pc_correct_math = 100*session_summary.n_correct_math / float(session_summary.n_math)
session_summary.math_per_list = session_summary.n_math / float(n_lists)
session_intr_events = intr_events[intr_events.session == session]
session_summary.n_pli = np.sum(session_intr_events.intrusion > 0)
session_summary.pc_pli = 100*session_summary.n_pli / float(n_sess_events)
session_summary.n_eli = np.sum(session_intr_events.intrusion == -1)
session_summary.pc_eli = 100*session_summary.n_eli / float(n_sess_events)
session_summary_array.append(session_summary)
self.pass_object('SESSION_DATA', session_data)
self.pass_object('session_summary_array', session_summary_array)
cumulative_summary = SessionSummary()
cumulative_summary.n_words = len(events)
cumulative_summary.n_correct_words = np.sum(events.recalled)
cumulative_summary.pc_correct_words = 100*cumulative_summary.n_correct_words / float(cumulative_summary.n_words)
cumulative_summary.irt_within_cat = sum(irt_within_cat) / len(irt_within_cat) if irt_within_cat else 0.0
cumulative_summary.irt_between_cat = sum(irt_between_cat) / len(irt_between_cat) if irt_between_cat else 0.0
positions = np.unique(events.serialpos)
prob_recall = np.empty_like(positions, dtype=float)
for i,pos in enumerate(positions):
pos_events = events[events.serialpos == pos]
prob_recall[i] = np.sum(pos_events.recalled) / float(len(pos_events))
cumulative_summary.prob_recall = prob_recall
prob_first_recall = first_recall_counter / float(total_list_counter)
cumulative_summary.prob_first_recall = prob_first_recall
cumulative_summary.n_correct_stim = cumulative_n_recalls_from_stim
cumulative_summary.n_total_stim = cumulative_n_items_from_stim
cumulative_summary.pc_from_stim = 100 * cumulative_n_recalls_from_stim / float(cumulative_n_items_from_stim)
cumulative_summary.n_correct_nonstim = cumulative_n_recalls_from_nonstim
cumulative_summary.n_total_nonstim = cumulative_n_items_from_nonstim
cumulative_summary.pc_from_nonstim = 100 * cumulative_n_recalls_from_nonstim / float(cumulative_n_items_from_nonstim)
cumulative_summary.n_stim_intr = cumulative_n_intr_from_stim
cumulative_summary.pc_from_stim_intr = 100 * cumulative_n_intr_from_stim / float(cumulative_n_items_from_stim)
cumulative_summary.n_nonstim_intr = cumulative_n_intr_from_nonstim
cumulative_summary.pc_from_nonstim_intr = 100 * cumulative_n_intr_from_nonstim / float(cumulative_n_items_from_nonstim)
cumulative_summary.chisqr,cumulative_summary.pvalue,_ = proportions_chisquare([cumulative_n_recalls_from_stim, cumulative_n_recalls_from_nonstim], [cumulative_n_items_from_stim, cumulative_n_items_from_nonstim])
cumulative_summary.chisqr_intr,cumulative_summary.pvalue_intr,_ = proportions_chisquare([cumulative_n_intr_from_stim, cumulative_n_intr_from_nonstim], [cumulative_n_items_from_stim, cumulative_n_items_from_nonstim])
if math_events is not None:
cumulative_summary.n_math = len(math_events)
cumulative_summary.n_correct_math = np.sum(math_events.iscorrect)
cumulative_summary.pc_correct_math = 100*cumulative_summary.n_correct_math / float(cumulative_summary.n_math)
cumulative_summary.math_per_list = cumulative_summary.n_math / float(total_list_counter)
cumulative_summary.n_pli = np.sum(intr_events.intrusion > 0)
cumulative_summary.pc_pli = 100*cumulative_summary.n_pli / float(len(events))
cumulative_summary.n_eli = np.sum(intr_events.intrusion == -1)
cumulative_summary.pc_eli = 100*cumulative_summary.n_eli / float(len(events))
self.pass_object('cumulative_summary', cumulative_summary)