def run(data, results_path, seed):
np.random.seed(seed)
results = {}
for name in sorted(data.keys()):
correct = data[name][data[name]['correct']]
results[name] = 1. / util.bootstrap_mean(1. / correct['time'])
results = pd.DataFrame.from_dict(results, orient='index')
results.index.name = 'model'
# these are out of order, so fix them
results = results.rename(columns={
'lower': 'upper',
'upper': 'lower'})
pth = results_path.joinpath(filename)
results.to_csv(pth)
with open(results_path.joinpath(texname), "w") as fh:
fh.write("%% AUTOMATICALLY GENERATED -- DO NOT EDIT!\n")
for model, stats in results.iterrows():
cmd = util.newcommand(
"%sTime" % model.capitalize(),
util.latex_msec.format(**dict(stats)))
fh.write(cmd)
return pth
def num_chance(df):
groups = df.groupby(['kappa0', 'stimulus'])['mass? correct']
alpha = (0.05 / len(groups.groups)) * 100
results = groups\
.apply(lambda x: util.bootstrap_mean(x, percentiles=[alpha]))\
.unstack(-1) <= 0.5
results = results.rename(columns={alpha: alpha / 100})
return results
def run(data, results_path, seed):
np.random.seed(seed)
keys = ['exp', 'expA', 'expB']
results = {}
for key in keys:
df = data[key]
times = df[df['correct']].groupby('trial')['time']
results[key] = times.apply(
lambda x: 1. / util.bootstrap_mean(1. / x))
results = pd.DataFrame.from_dict(results, orient='index')
results.index.name = 'model'
results.columns = pd.MultiIndex.from_tuples(
results.columns, names=['trial', 'stat'])
results = results.stack('trial')
pth = results_path.joinpath(filename)
results.to_csv(pth)
return pth
def run(data, results_path, seed):
np.random.seed(seed)
results = {}
for key, df in data.iteritems():
y = df[df['correct']].groupby(
['stimulus', 'modtheta', 'flipped'])['time']
results[key] = y.apply(
lambda x: 1. / util.bootstrap_mean(1. / x))
results = pd.DataFrame.from_dict(results)
results.index = pd.MultiIndex.from_tuples(
results.index, names=['stimulus', 'modtheta', 'flipped', 'stat'])
results.columns.name = 'model'
results = results.stack().unstack('stat')
results.columns.name = None
results = results.reset_index()
pth = results_path.joinpath(filename)
results.to_csv(pth)
return pth
def bootstrap_mean(df, **kwargs):
name, df = df
df.name = name
return util.bootstrap_mean(df, **kwargs)
