def test_count_feature_observed(self):
control_data = [
np.array([
['line', 'subject', 'tier2', 'tier3', 'tier1', 'tmin', 'tmax'],
['1', 'CF61WOM_47', 'sil', '?', '{noise}', '0', '1.375'],
[
'2', 'CF61WOM_47', 'clause1_exp-s1', 'exp',
'"allora parlaci un poco di te e della tua famiglia?" ',
'1.375', '4.046'
],
['3', 'CF61WOM_47', 'sil', '?', '(.)_exp', '4.046', '4.214'],
[
'4', 'CF61WOM_47', 'clause2_exp-s1', 'exp',
'"quanti componenti sono" ', '4.214', '5.469'
],
[
'5', 'CF61WOM_47', 'clause3_exp-s1', 'exp', '"e cosa fa"',
'5.469', '5.867'
], ['6', 'CF61WOM_47', 'sil', '?', '[]', '5.867', '6.137'],
[
'7', 'CF61WOM_47', 'clause1-s1', 'sub', 'quindi', '6.137',
'6.633'
]
])
]
depressed_data = [
np.array([
['line', 'subject', 'tier2', 'tier3', 'tier1', 'tmin', 'tmax'],
['1', 'PM33ATR_66', 'noise', '?', '{noise}', '0', '1.739'],
[
'2', 'PM33ATR_66', 'clause1_exp-s1', 'exp',
'"raccontamo un po\' come"', '1.739', '3.059'
], ['3', 'PM33ATR_66', 'sil', 'f', ':::e', '3.059', '3.375'],
['4', 'PM33ATR_66', 'fil', '?', ':m', '3.375', '4.046'],
[
'5', 'PM33ATR_66', 'clause1_exp-s2', '?',
'"hai passato quest\'ultima settimana"', '4.046', '5.727'
], ['6', 'PM33ATR_66', 'sil', '?', '(.)v', '5.727', '6.659'],
[
'7', 'PM33ATR_66', 'clause1-s1', 'sub',
"allora quest ultima settimana l'ho passata", '6.659',
'9.187'
]
])
]
feature = 'sil'
real_observed_f_count = [3, 2]
observed_f_count, expected_f_count = count_feature(
feature, control_data, depressed_data, 6.633, 9.187)
np.testing.assert_almost_equal(real_observed_f_count, observed_f_count,
2)
from counts.count_analysis import import_data
from counts.count_analysis import count_feature
from counts.count_analysis import return_chisquare
control_data, depressed_data, control_duration, depressed_duration = import_data()
observed, expected = count_feature('ove', control_data, depressed_data, control_duration, depressed_duration)
chisquare = return_chisquare(observed, expected)
print(chisquare)
features = [
'bac',
'cry',
'fil',
'lau',
'len',
'ove',
'sil'
]
pvalues = {}
control_data, depressed_data, control_duration, depressed_duration = import_data()
for feature in features:
obs, exp = count_feature(feature, control_data, depressed_data, control_duration, depressed_duration)
pvalue = return_chisquare(obs, exp)[1]
pvalues[feature] = pvalue
pvalues = {feature: pvalues[feature] for feature in pvalues.keys() if pvalues[feature] < 0.05}
sorted_pvalues = sorted(pvalues.items(), key=operator.itemgetter(1))
test_count = len(features)
thresholds = {}
for i in range(len(pvalues)):
threshold = ALPHA * (i+1) / test_count
thresholds[sorted_pvalues[i][0]] = threshold
for feature, pvalue in sorted_pvalues:
print('{0} {1} -- threshold: {2}'.format(feature, pvalue, thresholds[feature]))