DiffQue/common2.py at master · deepak0004/DiffQue · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
import pickle
from Question import Question
from random import randint
from sklearn.metrics import f1_score
from sklearn import metrics
from hits_nx import hits_nx
import networkx as nx
from removenewline import RemoveNewLine
from os import system
from sys import argv
import sys

questions = dict()

user_best_answers_question_id = dict()
user_non_best_answers_question_id = dict()

computeF1 = True


#Global setting for dataset. Note that changing the dataset here is enough AFAIK
dataset = "old" #choose "math", "old" or "new"
if len(argv) >= 2:
	dataset = argv[1]


print "Dataset in use is",dataset

sys.stdout.flush()

if dataset == "old":
	with open("dump2/old_so/old_so_questions4.dump", "rb") as fp:   # Unpickling
		questions = pickle.load(fp)
	with open("dump2/old_so/old_so_user_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_best_answers_question_id = pickle.load(fp)
	with open("dump2/old_so/old_so_user_non_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_non_best_answers_question_id = pickle.load(fp)
	with open("dump2/old_so/old_so_user_questions4.dump", "rb") as fp:   # Unpickling
		user_questions = pickle.load(fp)
elif dataset == "new":
	with open("dump2/new_so/new_so_questions4.dump", "rb") as fp:   # Unpickling
		questions = pickle.load(fp)
	with open("dump2/new_so/new_so_user_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_best_answers_question_id = pickle.load(fp)
	with open("dump2/new_so/new_so_user_non_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_non_best_answers_question_id = pickle.load(fp)
	with open("dump2/new_so/new_so_user_questions4.dump", "rb") as fp:   # Unpickling
		user_questions = pickle.load(fp)
elif dataset == "math":
	with open("dump2/math/math_questions4.dump", "rb") as fp:   # Unpickling
		questions = pickle.load(fp)
	with open("dump2/math/math_user_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_best_answers_question_id = pickle.load(fp)
	with open("dump2/math/math_user_non_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_non_best_answers_question_id = pickle.load(fp)
	with open("dump2/math/math_user_questions4.dump", "rb") as fp:   # Unpickling
		user_questions = pickle.load(fp)
elif dataset == "mid":
	with open("dump2/mid_so/mid_so_questions4.dump", "rb") as fp:   # Unpickling
		questions = pickle.load(fp)
	with open("dump2/mid_so/mid_so_user_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_best_answers_question_id = pickle.load(fp)
	with open("dump2/mid_so/mid_so_user_non_best_answers_question_id4.dump", "rb") as fp:   # Unpickling
		user_non_best_answers_question_id = pickle.load(fp)
	with open("dump2/mid_so/mid_so_user_questions4.dump", "rb") as fp:   # Unpickling
		user_questions = pickle.load(fp)
else:
	print "Dataset invalid"
	assert(False)


# with open("dump2/answer_texts_by_question_id4.dump", "rb") as fp:   # Unpickling
# 	answer_texts_by_question_id4 = pickle.load(fp)


def computeAccuracyAndF1AndAUC(ratings):
	global dataset
	fin_pairs = open("questions/ground_truth/" + dataset+ ".txt") #Kindly put all the three files in this directory, named as per the possible values of dataset writen at its declaration
	print "Same rating pairs(if any) :"
	correct = 0.0
	incorrect = 0.0
	rejected1 = 0
	rejected2 = 0
	y_true = []
	y_pred = []
	for line in fin_pairs:
		if len(line.split(" ")) <= 2:
			continue
		line = line.rstrip()
		id1 = line.split(" ")[0]
		id2 = line.split(" ")[1]
		better_id = line.split(" ")[2]
		rating1 = None
		rating2 = None
		try:
			rating1 = ratings[int(id1)]
		except:
			rejected1 += 1
			continue
		try:
			rating2 = ratings[int(id2)]
		except:
			rejected1 += 1
			continue
		if better_id == id1:
			y_true.append(0)
		elif better_id == id2:
			y_true.append(1)
		else:
			assert(False)
		if rating1 > rating2:
			y_pred.append(0)
		elif rating2 > rating1:
			y_pred.append(1)
		else:
			r = randint(0, 1) #Randomly predict more difficult question
			y_pred.append(r)
		if rating1 > rating2 and better_id == id1:
			correct += 1
		elif rating2 > rating1 and better_id == id2:
			correct += 1
		else:
			# temp_file.write(id1+" "+id2+"\n")
			# print id1,id2
			if rating1 == rating2:
				print id1, id2
				rejected2 += 1
				r = randint(1, 2) #Randomly predict more difficult question
				if r == 1 and better_id == id1:
					correct += 1
				elif r == 2 and better_id == id2:
					correct += 1
				else:
					incorrect += 1
			else:
				# print id1, id2, better_id
				incorrect += 1
			# if rating1 == rating2:
			# 	rejected2 += 1
			# 	print "Rating same for",id1,id2

	print "Rejected(absentees)", rejected1, "PseudoRejected(same rating)", rejected2
	print "Accuracy", correct/(correct + incorrect), "Analyzed",correct + incorrect
	print "Note that same ratings are randomly assigned a different difficulty!"
	print "F1 Score:", f1_score(y_true, y_pred, average='macro')
	fpr, tpr, thresholds = metrics.roc_curve(y_true, y_pred)
	print "AUC:", metrics.auc(fpr, tpr)
	print "Dataset is",dataset
	fin_pairs.close()

def computeAccuracy(ratings):
	global dataset
	print "Dataset in use is",dataset
	if computeF1:
		print "Redirecting to compute both F1 and accuracy"
		computeAccuracyAndF1AndAUC(ratings)
		return
	fin_pairs = open("questions/ques_pairs_with_groundtruth.txt")
	print "Same rating pairs(if any) :"
	correct = 0.0
	incorrect = 0.0
	rejected1 = 0
	rejected2 = 0
	for line in fin_pairs:
		if len(line.split(" ")) <= 2:
			continue
		line = line.rstrip()
		id1 = line.split(" ")[0]
		id2 = line.split(" ")[1]
		better_id = line.split(" ")[2]
		rating1 = None
		rating2 = None
		try:
			rating1 = ratings[int(id1)]
		except:
			rejected1 += 1
			continue
		try:
			rating2 = ratings[int(id2)]
		except:
			rejected1 += 1
			continue
		if rating1 > rating2 and better_id == id1:
			correct += 1
		elif rating2 > rating1 and better_id == id2:
			correct += 1
		else:
			# temp_file.write(id1+" "+id2+"\n")
			# print id1,id2
			if rating1 == rating2:
				print id1, id2
				rejected2 += 1
				r = randint(1, 2) #Randomly predict more difficult question
				if r == 1 and better_id == id1:
					correct += 1
				elif r == 2 and better_id == id2:
					correct += 1
				else:
					incorrect += 1
			else:
				# print id1, id2, better_id
				incorrect += 1
			# if rating1 == rating2:
			# 	rejected2 += 1
			# 	print "Rating same for",id1,id2

	print "Accuracy", correct/(correct + incorrect), "Analyzed",correct + incorrect, "Rejected(absentees)", rejected1, "PseudoRejected(same rating)", rejected2
	print "Note that same ratings are randomly assigned a different difficulty!"
	fin_pairs.close()