def askCrowd(data, mode): #for asking a sample from crowd
cp = CrowsDis("../data/ProtonBeamCrowddata.txt")
cp.proportion()
cp.voting()
feature_dict = {
i: label
for i, label in zip(range(3), ('pmid', 'text', 'label'))
}
crowddata = (data[[0, 1]].values).tolist()
label = []
if mode == 'all':
for i in range(0, len(data["pmid"])):
label.append(cp.getvotes()[data['pmid'][i]])
if mode == '1':
pidval = data['pmid'].values
for i in range(0, len(data["pmid"])):
label.append(cp.getcrowdvotequestion1()[pidval[i]])
crowdlabel = (np.array([label]).T).tolist()
data = np.array([t1 + t2 for t1, t2 in zip(crowddata, crowdlabel)])
columns = [l for i, l in sorted(feature_dict.items())]
return pd.DataFrame(data, columns=columns) #make dataframe from M sample
def wallace(pid, y_real):
cp = CrowsDis("../data/ProtonBeamCrowddata.txt")
cp.proportion()
for i in range(1, 6):
print i
label = []
cp.wallaceExpriment(i)
for item in pid:
label.append(cp.getvotes()[item])
cm = confusion_matrix(y_real, label).flatten()
sensitivity, specificity, precision, loss = compute_measures(*cm)
print sensitivity
print specificity
def getcustomizevoting(crowdlist, pid):
result = defaultdict(list)
cp = CrowsDis("../data/ProtonBeamCrowddata.txt")
cp.proportion()
combinationcrowd = []
for i in xrange(4, len(crowdlist)):
combinationcrowd.append(list(itertools.combinations(crowdlist, i)))
for indexgroup in range(len(combinationcrowd)): #
for item in range(len(combinationcrowd[indexgroup])):
label = [] # convert crowd votes dict to ordered labels
cp.customizevoting(len(crowdlist),
list(combinationcrowd[indexgroup][item]))
print cp.getvotes()
for item in pid:
label.append(
cp.getvotes()[item]) # to be add lables in the same order
result[indexgroup].append(label)
return result