-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathevaluate.py
205 lines (177 loc) · 8.28 KB
/
evaluate.py
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
import numpy
import os
import sklearn
import utils as ut
from sklearn.metrics.cluster import normalized_mutual_info_score, completeness_score
from sklearn.metrics.cluster import v_measure_score, adjusted_rand_score, homogeneity_score
from sklearn.metrics.cluster import contingency_matrix, adjusted_mutual_info_score,fowlkes_mallows_score
from sklearn.metrics import accuracy_score
import ntpath
import time
from collections import Counter
import warnings
warnings.simplefilter("ignore")
def purity_score(y_true, y_pred):
# compute contingency matrix (also called confusion matrix)
contingency_matrix = sklearn.metrics.cluster.contingency_matrix(y_true, y_pred)
# return purity
return numpy.sum(numpy.amax(contingency_matrix, axis=0)) / numpy.sum(contingency_matrix)
def loadResultFile(resFile):
# print("loading file: ", resFile)
clusDocMap={}
clusIDs = []
docCluster = {}
with open(resFile) as input:
line = input.readline()
while line:
docIdPredClusId = line.strip().split(' ') # [documentID, ClusterID]
currentDocID = docIdPredClusId[0]
currentCluID = docIdPredClusId[1]
docCluster[currentDocID] = currentCluID
if currentCluID not in clusIDs:
clusIDs.append(currentCluID)
clusDocMap[currentCluID] = []
clusterDocs = clusDocMap[currentCluID]
clusterDocs.append(currentDocID)
line = input.readline()
input.close()
return clusDocMap, clusIDs, docCluster
#End Function
def accuracy(clusDocMap, docsClassMap):
doc_cluster_map_cluster_to_class_conversion = {}
for clussID, doccIDs in clusDocMap.items():
count_instance_of_class = {}
max_instance = 0
max_class = ''
tmp_docsClassMap = docsClassMap
# tmp_docsClassMap = {} # FOR MSTreamF
# for d_id, c_id in docsClassMap.items(): # FOR MSTreamF
# tmp_docsClassMap[str(int(d_id))] = c_id
for doccID in doccIDs:
class_name = tmp_docsClassMap[doccID]
instance_total = count_instance_of_class.get(class_name, 0)
count_instance_of_class[class_name] = (instance_total + 1)
if (instance_total + 1) > max_instance:
max_instance = (instance_total + 1)
max_class = class_name
for doccID in doccIDs:
doc_cluster_map_cluster_to_class_conversion[doccID] = class_name
return doc_cluster_map_cluster_to_class_conversion
def evaluate_file(file_path, docsClassMap, classDocMap, stats_dir, file_name, summary_file = None):
clusDocMap, clusIDs, docCluster = loadResultFile(file_path)
# Updated for advance evaluation
originalArray = []
predictedArray = []
predictedArrayForAccuracy = []
doc_class_class_map = accuracy(clusDocMap, docsClassMap)
for docId, classID in docsClassMap.items():
# predictedClusterId = docCluster[str(int(docId))] # FOR MSTreamF
predictedClusterId = docCluster[docId]
originalArray.append(classID)
predictedArray.append(predictedClusterId)
predictedArrayForAccuracy.append(doc_class_class_map[docId])
# predictedArrayForAccuracy.append(doc_class_class_map[str(int(docId))]) # FOR MSTreamF
nmi = normalized_mutual_info_score(originalArray, predictedArray)
ari = adjusted_rand_score(originalArray, predictedArray)
v_score = v_measure_score(originalArray, predictedArray)
com_score = completeness_score(originalArray, predictedArray)
homo = homogeneity_score(originalArray, predictedArray)
ami = adjusted_mutual_info_score(originalArray, predictedArray)
fms = fowlkes_mallows_score(originalArray, predictedArray)
purity = purity_score(originalArray, predictedArray)
inverse_purity = purity_score( predictedArray, originalArray)
acc = accuracy_score(originalArray, predictedArrayForAccuracy)
# --- updated code enede
# tool = NMI(classDocMap, clusDocMap, docsClassMap)
# nmi = tool.calculate()
temp = file_path + "\t "
temp = temp + str(nmi) + " \t "
temp = temp + str(ari) + " \t "
temp = temp + str(v_score) + "\t "
temp = temp + str(com_score) + "\t "
temp = temp + str(homo) + "\t"
temp = temp + str(ami) + " \t "
temp = temp + str(fms) + " \t "
temp = temp + str(purity) + " \t"
temp = temp + str(acc) + "\t"
temp = temp + str(inverse_purity) + " \t"
temp = temp + str(clusDocMap.__len__()) + " \t"
temp = temp + str(classDocMap.__len__())
console = temp.replace("_ALPHA", "\t").replace("_BETA", "\t").replace(".txt", "")
print(console)
if summary_file != None:
summary_file.write(temp)
summary_file.write("\n")
statFile = file_name.replace(".txt", "_STATISTICS.txt")
f = open(stats_dir + statFile, "w")
f.write(temp)
f.write("\n")
for clusterID, documents in clusDocMap.items():
temp = str(clusterID) + " \t" + str(documents.__len__()) + " \n"
f.write(temp)
f.close()
return nmi
def evaluate(classDocMap, docsClassMap, prediction_dir, stats_dir, highest_nmi_by_dir = True):
ath = stats_dir + "/"
try:
os.makedirs(stats_dir)
except:
print(stats_dir, " already exists")
total_files = 0
header = "File\t NMI \t ARI \tV_Measure \t Completeness\t Homogeneity\t AMI \tFMS \tpurity \t Accuracy \t I_puri \tClusters \t classes \n"
print(header)
isFile = os.path.isfile(prediction_dir)
highest_nmi = 0.0
highest_nmi_file = ""
if isFile:
highest_nmi = evaluate_file(prediction_dir, docsClassMap, classDocMap, stats_dir, ntpath.basename(prediction_dir))
else:
for r, directories, list_of_files in os.walk(prediction_dir): # getting predicted clusters and documents
summary = r +"/#summary-"+str(time.time())
summary_file = open(summary,"w")
summary_file.write(header)
dir_highest_nmi = 0.0
dir_highest_nmi_file = ""
for file in list_of_files:
if ("#summary" in file):
continue
nmi = evaluate_file(r + "/" + file,docsClassMap, classDocMap, stats_dir, file, summary_file=summary_file)
total_files+=1
if nmi > dir_highest_nmi:
dir_highest_nmi = nmi
dir_highest_nmi_file = (r + "/" + file)
if highest_nmi_by_dir == True:
summary_file.write("---------Directory Highest NMI------------- \n")
summary_file.write(str(dir_highest_nmi_file)+"\t "+str(dir_highest_nmi)+"\n")
print("---------Directory Highest NMI-------------")
print(str(dir_highest_nmi_file) + "\t " + str(dir_highest_nmi))
print("-------------------------------------------")
if highest_nmi < dir_highest_nmi:
highest_nmi = dir_highest_nmi
highest_nmi_file = dir_highest_nmi_file
summary_file.close()
print("TOTAL FILES: ",total_files)
return highest_nmi, highest_nmi_file
def evaluate_results(dataset, prediction_dir, stats_dir):
# if (not os.path.exists(stats_dir)):
# os.makedirs(stats_dir)
classDocMap, docsClassMap = ut.loadOrigialDocClassLabels(dataset)
highest_nmi, highest_nmi_file = evaluate(classDocMap,docsClassMap,prediction_dir, stats_dir)
print("----------- HIGHEST NMI --------")
print(highest_nmi_file," ", highest_nmi)
if __name__ == '__main__':
dataDir = "data/"
# outputPath = "F:/PhD/Coding/OSDM/venv/test/"
# outputPath = "F:/PhD/Coding/DTM/dtm-master/dtm-master/dtm/News-T-N/model_run/lda-seq/clustering_results.txt"
# outputPath = "result/MStreamF/News-T-NK0iterNum1SampleNum1alpha0.03beta0.03BatchNum16BatchSaved2.txt"
statPath = "stats/"
# dataset = dataDir+"News11104"
# dataset = dataDir+"News-T11104"
# dataset = dataDir+"News"
# dataset = dataDir+"reuters9445"
# dataset = dataDir+"reuters21578"
# dataset = dataDir+"reuters21578-T"
# dataset = dataDir+"Tweets"
# dataset = dataDir+"Tweets-T"
dataset = dataDir+"News-T-N"
evaluate_results(dataset,outputPath,statPath)