for alpha in alpha_list:
positive_count = sum([1 for x in X_positive if x >= alpha])
positive_rate = positive_count / float(len(X_positive))
negative_count = sum([1 for x in X_negative if x >= alpha])
negative_rate = negative_count / float(len(X_negative))
roc_X.append(negative_rate)
roc_Y.append(positive_rate)
#print roc_X
plt.plot(roc_X, roc_Y, color="green")
classify = IterativeEM()
classify.__classify__(vote_list, 2)
estimates = classify.__getEstimates__()
X_positive = []
X_negative = []
for subject_index, zooniverse_id in enumerate(big_subjectList):
probability = estimates[subject_index]
wreness_condor = gold_condor[subject_index]
if wreness_condor == 0:
X_negative.append(probability)
else:
X_positive.append(probability)
alpha_list = X_negative[:]
alpha_list.extend(X_positive)
alpha_list.sort()
negative_count = sum([1 for x in X_negative if x >= alpha])
negative_rate = negative_count/float(len(X_negative))
roc_X.append(negative_rate)
roc_Y.append(positive_rate)
#print roc_X
plt.plot(roc_X,roc_Y,color="green")
classify = IterativeEM()
classify.__classify__(vote_list,2)
estimates = classify.__getEstimates__()
X_positive = []
X_negative = []
for subject_index,zooniverse_id in enumerate(big_subjectList):
probability = estimates[subject_index]
wreness_condor = gold_condor[subject_index]
if wreness_condor == 0:
X_negative.append(probability)
else:
X_positive.append(probability)
alpha_list = X_negative[:]
alpha_list.extend(X_positive)
alpha_list.sort()
votes.append((user_index, animal_index, 1))
if not (animal_index in animal_votes):
animal_votes[animal_index] = [1]
else:
animal_votes[animal_index].append(1)
else:
votes.append((user_index, animal_index, 0))
if not (animal_index in animal_votes):
animal_votes[animal_index] = [0]
else:
animal_votes[animal_index].append(0)
print "=====---"
#print votes
classify = IterativeEM()
classify.__classify__(votes)
most_likely = classify.__getMostLikely__()
estimates = classify.__getEstimates__()
X = []
Y = []
X2 = []
Y2 = []
#for subject_index,zooniverse_id in enumerate(big_subjectList):
for ii in range(animal_index):
x = np.mean(animal_votes[ii])
y = estimates[ii][1]
X.append(x)
votes.append((user_index, subject_index, 0))
if not (zooniverse_id in subject_vote):
subject_vote[zooniverse_id] = [0]
else:
subject_vote[zooniverse_id].append(0)
except ValueError:
votes.append((user_index, subject_index, 0))
if not (zooniverse_id in subject_vote):
subject_vote[zooniverse_id] = [0]
else:
subject_vote[zooniverse_id].append(0)
print "=====---"
classify = IterativeEM()
classify.__classify__(votes)
most_likely = classify.__getMostLikely__()
estimates = classify.__getEstimates__()
X = []
Y = []
X2 = []
Y2 = []
for subject_index, zooniverse_id in enumerate(big_subjectList):
subject = subject_collection.find_one({"zooniverse_id": zooniverse_id})
if zooniverse_id in subject_vote:
x = np.mean(subject_vote[zooniverse_id])
y = estimates[subject_index][1]
for animal,most_likely in zip(animal_votes,plurality_vote):
for reported in animal:
confusion[reported][most_likely] += 1
confusion = [[max(c/sum(row),0.005) for c in row] for row in confusion]
#print (confusion[3][0],confusion[3][3])
confusion_list= individual_confusion(votes,plurality_vote)
#for c in confusion_list:
# print c
#assert False
#print votes
classify = IterativeEM()
classify.__classify__(votes,len(animals),confusion_list)
most_likely = classify.__getMostLikely__()
estimates = classify.__getEstimates__()
X = []
Y = []
X2 = []
Y2 = []
#print animals
#for subject_index,zooniverse_id in enumerate(big_subjectList):
for ii,votes in enumerate(animal_votes):
percentage = [sum([1 for v in votes if v == a_t])/float(len(votes)) for a_t in range(len(animals))]
EM_percentage = estimates[ii]
for animal, most_likely in zip(animal_votes, plurality_vote):
for reported in animal:
confusion[reported][most_likely] += 1
confusion = [[max(c / sum(row), 0.005) for c in row] for row in confusion]
#print (confusion[3][0],confusion[3][3])
confusion_list = individual_confusion(votes, plurality_vote)
#for c in confusion_list:
# print c
#assert False
#print votes
classify = IterativeEM()
classify.__classify__(votes, len(animals), confusion_list)
most_likely = classify.__getMostLikely__()
estimates = classify.__getEstimates__()
X = []
Y = []
X2 = []
Y2 = []
#print animals
#for subject_index,zooniverse_id in enumerate(big_subjectList):
for ii, votes in enumerate(animal_votes):
percentage = [
sum([1 for v in votes if v == a_t]) / float(len(votes))
for a_t in range(len(animals))
]
# #print (confusion[3][0],confusion[3][3])
#
# X,Y= individual_confusion(votes,plurality_vote)
# print X
# print Y
# plt.hist(X,bins=20)
# # plt.plot(X,Y,'.',color="blue")
# # plt.plot([X_avg,],[Y_avg,],'.',color="green")
# # plt.xlim((-0.05,1.05))
# # plt.ylim((0.95,1.05))
# plt.show()
# assert False
#print votes
classify = IterativeEM()
classify.__classify__(votes,len(animals),confusion,gold_values=gold_values)
most_likely = classify.__getMostLikely__()
estimates = classify.__getEstimates__()
X = []
Y = []
X2 = []
Y2 = []
#print animals
#for subject_index,zooniverse_id in enumerate(big_subjectList):
for ii,votes in enumerate(animal_votes):
percentage = [sum([1 for v in votes if v == a_t])/float(len(votes)) for a_t in range(len(animals))]
EM_percentage = estimates[ii]