def staircase(self):
df = self._get_staircase()
agg_thres = stats.aggregate(df,
values='thres',
rows='session',
unstack=True)
agg = stats.aggregate(df,
values='oridiff',
subplots='session',
rows='trialno',
cols='pos')
if self.rp['plot'] or self.rp['saveplot']:
plt = plot.Plot(nrows=2, ncols=2, sharex=True, sharey=True)
ax = plt.plot(agg, kind='line', ylabel='step')
num_trials = len(agg.columns.levels[1])
for pno, (sno, thres) in enumerate(agg_thres.mean().iteritems()):
#import pdb; pdb.set_trace()
ax = plt.get_ax(pno)
ax.set_title('session %d' % sno)
ax.axhline(y=thres, color='.2', ls='--', lw=2, marker='None')
print thres
ax.text(num_trials - 15,
thres + .5,
'average threshold = %.2f' % thres,
fontsize=8)
print agg_thres
plt.hide_plots(3)
if self.rp['plot']: plt.show()
if self.rp['saveplot']:
plt.savefig(PATHS['analysis'] + '%s.svg' % self.info['subjid'])
return agg_thres
def run(self):
df = self._get_staircase()
agg_thres = stats.aggregate(df, values='thres', rows='pos', unstack=True)
agg = stats.aggregate(df, values='oridiff', rows='trialno', cols='pos')
plt = plot.Plot()
ax = plt.plot(agg, kind='line', ylabel='step')
num_trials = len(agg.columns.levels[1])
thres = agg_thres.mean(1).values[0]
ax.axhline(y=thres, color = '.2', ls='--', lw=2, marker = 'None')
print agg_thres
ax.text(num_trials - num_trials/3., thres+.5,
'average threshold = %.2f' % thres, fontsize=8)
plt.show()
return agg_thres
def get_agg(self, df, kind=None):
if kind == 'matrix':
agg = stats.aggregate(df, values='subj_resp', yerr='subjid',
subplots='roi', rows='stim1.cond',
cols='stim2.cond')
elif self.rp['method'] in ['corr', 'svm']:
df['same_diff'] = 'between'
df['same_diff'][df['stim1.cond'] == df['stim2.cond']] = 'within'
agg = stats.aggregate(df, values='subj_resp', yerr='subjid',
rows='roi', cols='same_diff')
else:
agg = stats.aggregate(df, values='subj_resp', yerr='subjid',
rows='roi', cols='cond')
return agg
def run(self):
pattern = PATHS['data'] + '%s.csv'
df = self.exp.get_behav_df(pattern=pattern)
agg_acc = stats.accuracy(df,
cols='context',
values='accuracy',
yerr='subjid',
order='sorted')
agg_rt = stats.aggregate(df[df.accuracy == 'correct'],
cols='context',
values='rt',
yerr='subjid',
order='sorted')
plt = plot.Plot(ncols=2)
if len(df.subjid.unique()) == 1:
kind = 'bar'
else:
kind = 'bean'
plt.plot(agg_acc, kind=kind, title='accuracy', ylabel='% correct')
plt.plot(agg_rt, kind=kind, title='response time', ylabel='seconds')
print agg_acc
print agg_rt
plt.show()
def behav(self):
pattern = PATHS['data'] + '%s.csv'
df = self.exp.get_behav_df(pattern=pattern)
agg_acc = stats.aggregate(df, cols='context', values='accuracy',
value_filter='Correct', yerr='subjID')
agg_rt = stats.aggregate(df[df.accuracy=='Correct'], cols='context',
values='rt', yerr='subjID')
if self.runParams['plot'] or self.runParams['saveplot']:
plt = plot.Plot(ncols=2)
plt.plot(agg_acc, kind='bar')
plt.plot(agg_rt, kind='bar')
if self.runParams['plot']: plt.show()
if self.runParams['saveplot']:
plt.savefig(PATHS['analysis'] + 'behav.svg')
def thresholds(self):
df = self._get_staircase()
agg_thres = stats.aggregate(df, values='thres', cols='pos',
yerr='subjid')
plt = plot.Plot()
plt.plot(agg_thres, kind='bar', title='orientation thresholds',
ylabel='orientation threshold, deg')
print agg_thres.mean()
plt.show()
return agg_thres
def run(self):
pattern = PATHS['data'] + '%s.csv'
df = self.exp.get_behav_df(pattern=pattern)
agg_acc = stats.accuracy(df, cols='context', values='accuracy',
yerr='subjid', order='sorted')
agg_rt = stats.aggregate(df[df.accuracy=='correct'], cols='context',
values='rt', yerr='subjid', order='sorted')
plt = plot.Plot(ncols=2)
if len(df.subjid.unique()) == 1:
kind = 'bar'
else:
kind = 'bean'
plt.plot(agg_acc, kind=kind, title='accuracy', ylabel='% correct')
plt.plot(agg_rt, kind=kind, title='response time', ylabel='seconds')
print agg_acc
print agg_rt
plt.show()
nsampl = 10
k = n*nsampl
df = stats.df_fromdict(
[('subplots', ['session1', 'session2']),
('cond', [1, 2]),
('name', ['condition one', 'condition two']),
('levels', ['small', 'medium', 'large']),
('subjID', ['subj%d' % (i+1) for i in range(n)])],
repeat=nsampl
)
df['rt'] = 0.5
df.rt[df.cond==1] = np.random.random(12*k) * 1.2
df.rt[df.cond==2] = np.random.random(12*k)
#df['accuracy'] = ['correct','correct','incorrect','incorrect']*k*9
agg = stats.aggregate(df, subplots='subplots', rows='levels',
cols='cond', yerr='subjID', values='rt')
plt1 = plot.Plot(figsize=(6,3))
plt1.plot(agg, kind='bar')
plt1.tight_layout()
plt2 = plot.Plot(figsize=(6,3))
plt2.plot(agg, kind='line')
plt2.tight_layout()
plt3 = plot.Plot(figsize=(8,4))
plt3.plot(agg, kind='bean')
plt3.tight_layout()
agg = stats.aggregate(df, subplots='subplots', rows='subjID',
cols='name', values='rt')
plt4 = plot.Plot(figsize=(6,3))
nsampl = 10
k = n * nsampl
df = stats.df_fromdict([('subplots', ['session1', 'session2']),
('cond', [1, 2]),
('name', ['condition one', 'condition two']),
('levels', ['small', 'medium', 'large']),
('subjID', ['subj%d' % (i + 1) for i in range(n)])],
repeat=nsampl)
df['rt'] = 0.5
df.rt[df.cond == 1] = np.random.random(12 * k) * 1.2
df.rt[df.cond == 2] = np.random.random(12 * k)
#df['accuracy'] = ['correct','correct','incorrect','incorrect']*k*9
agg = stats.aggregate(df,
subplots='subplots',
rows='levels',
cols='cond',
yerr='subjID',
values='rt')
plt1 = plot.Plot(figsize=(6, 3))
plt1.plot(agg, kind='bar')
plt1.tight_layout()
plt2 = plot.Plot(figsize=(6, 3))
plt2.plot(agg, kind='line')
plt2.tight_layout()
plt3 = plot.Plot(figsize=(8, 4))
plt3.plot(agg, kind='bean')
plt3.tight_layout()
agg = stats.aggregate(df,
