-
Notifications
You must be signed in to change notification settings - Fork 0
/
word_net_create_2.py
176 lines (121 loc) · 6.43 KB
/
word_net_create_2.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
from tqdm import tqdm
import numpy as np
import pandas as pd
tqdm.pandas()
import os
os.chdir("/Users/chaofanzhai/UMN/Project_memory/data/find_dynamic_relationship")
df = pd.read_csv('tmp_word_net_dataset.csv')
# 考研单词书， sep 21 - nov 1
# first time study -word
# # uid :52779, # spell: 2023
df['first_response'] = df['first_response'].map(lambda x: {1:1,2:0,3:0,4:1}[x]) # 1:认识,2:模糊,3:忘记,4:熟知
df_gp = df.groupby(['pts', 'user_id'])[['spelling', 'first_response']].agg(lambda x: list(x))
df_gp = df_gp.reset_index(drop=False)
df_gp.columns = ['pts', 'user_id', 'word_list_bydate', 'first_response_list']
df_gp.to_csv('tmp_word_net_dataset-gp.tsv',sep='\t',index=None)
## filter
temp = df_gp.groupby(["user_id"]).count()["pts"]
temp = temp.reset_index(drop=False)
temp.groupby(['pts']).count()
temp.describe()
wordset = tuple(set(df.spelling.values))
print("num of words", len(wordset))
def get_word_vector(x):
word_list_bydate=x[0]
first_response_list=x[1]
dic = dict(zip(word_list_bydate,first_response_list))
vector = []
for word in wordset:
if word in dic.keys():
vector.append(dic[word])
else:
vector.append(np.nan)
return vector
print('generate word vector')
#df_gp['vector']= df_gp[['word_list_bydate','first_response_list']].progress_apply(get_word_vector, axis= 1)
df_gp['vector']= df_gp[['word_list_bydate','first_response_list']].apply(get_word_vector, axis= 1)
# using a dataframe to store the bag of words
print("generate word matrix")
X=[]
for i in tqdm(df_gp['vector'].values) :
X.append(i)
matrix = np.asarray(X)
matrix_pd = pd.DataFrame(matrix) # each row is user-date level observation. each column is a word.
m = matrix_pd.head(10000)
# check the shared google doc sktech
def get_pair_frequency(matrix_pd: pd.DataFrame):
Z = matrix_pd==1
Z = Z.astype(int) # 记住=1，没记、没学=0
L = matrix_pd.notna().astype(int) # 有学=1, 没学=0
## map reduce ,
return L.T.dot(Z) , L.T.dot(L) # L.T.dot(Z): row:有没有学 col:有没有记住
#remember_pair_frequency_matrix, study_pair_frequency_matrix = get_pair_frequency(matrix_pd)
remember_pair_frequency_matrix, study_pair_frequency_matrix = get_pair_frequency(m)
## -- 出现-》记住
def find_frequent_item_pair(supp_threshold,cond_threshold,lift_threshold, wordset) :
# update the significance ,
length = matrix_pd.shape[0]
pair_frequency_matrix = study_pair_frequency_matrix/length #同时出现的次数/所有记录的长度
cond_matrix = remember_pair_frequency_matrix/study_pair_frequency_matrix #同时记住的次数/同时出现的次数
print(cond_matrix)
diag = np.diag(cond_matrix) # diag: X 记住/X 出现
length = diag.shape[0]
lift_matrix = cond_matrix/diag.reshape(1,length) #（ 同时记住的次数/同时出现的次数）/（X记住的次数/x出现的次数）
print(lift_matrix)
x = pair_frequency_matrix>supp_threshold
y = cond_matrix>cond_threshold
z = lift_matrix>lift_threshold
S = x*y*z # S satisfy the both conditions
#M= np.tril(S,-1) # get the Lower triangle of an array.
loc = np.where(S==True)
pairs ={}
for i, pair_x in enumerate(tqdm(loc[0])):
pair_y = loc[1][i]
pairs[(wordset[pair_x],wordset[pair_y])]= (pair_frequency_matrix.loc[pair_x,pair_y], cond_matrix.loc[pair_x,pair_y], lift_matrix.loc[pair_x,pair_y])
return pairs
def find_frequent_item_pair_2(matrix_pd, supp_thres1,cond_thres1,lift_thres1,supp_thres2,cond_thres2,lift_thres2, wordset) :
# update the significance ,
supp_matrix1 = remember_pair_frequency_matrix/study_pair_frequency_matrix #同时记住的次数/同时出现的次数
supp_matrix2= remember_pair_frequency_matrix/ matrix_pd.shape[0] #同时记住的次数/所有记录的长度
diag = np.diag(remember_pair_frequency_matrix)
length = diag.shape[0]
diag = diag.reshape(1,length) # row #=1, column # = length
diag_T = diag.reshape(length,1)
diag_study = np.diag(study_pair_frequency_matrix).reshape(1,length)
diag_study_T = diag_study.reshape(length,1)
cond_matrix1 = remember_pair_frequency_matrix/ diag #同时记住的次数/ X记住的次数
cond_matrix2 = remember_pair_frequency_matrix/ diag_study #同时记住的次数/ X出现的次数
lift_matrix1 = (remember_pair_frequency_matrix/diag_study)/ (diag_T/diag_study_T) #（同时记住的次数/X出现的次数 ）/（Y记住的次数/Y出现的次数）
lift_matrix2 = (remember_pair_frequency_matrix/study_pair_frequency_matrix)/ (diag_T/diag_study_T) #（同时记住的次数/同时出现的次数 ）/（Y记住的次数/Y出现的次数）
# lift_matrix3 = (remember_pair_frequency_matrix/study_pair_frequency_matrix)/ (diag_T/diag_study_T) #（Y记住的次数/同时出现的次数 ）/（Y记住的次数/Y出现的次数）
x1 = supp_matrix1>supp_thres1
y1 = cond_matrix1>cond_thres1
z1 = lift_matrix1>lift_thres1
x2 = supp_matrix2>supp_thres2
y2 = cond_matrix2>cond_thres2
z2= lift_matrix2>lift_thres2
S = x1*y1*z1*x2*y2*z2 # S satisfy the both conditions
#M= np.tril(S,-1) # get the Lower triangle of an array.
loc = np.where(S==True)
pairs ={}
for i, pair_x in enumerate(tqdm(loc[0])):
pair_y = loc[1][i]
pairs[(wordset[pair_x],wordset[pair_y])]= (supp_matrix1.loc[pair_x,pair_y],supp_matrix2.loc[pair_x,pair_y], cond_matrix1.loc[pair_x,pair_y], cond_matrix2.loc[pair_x,pair_y], lift_matrix1.loc[pair_x,pair_y],lift_matrix2.loc[pair_x,pair_y])
return pairs
supp_thres1,cond_thres1,lift_thres1,supp_thres2,cond_thres2,lift_thres2=0,0,0,0,0,0
frequent_pairs =find_frequent_item_pair_2(matrix_pd, supp_thres1,cond_thres1,lift_thres1,supp_thres2,cond_thres2,lift_thres2, wordset)
d = {}
for i, key in enumerate(tqdm(frequent_pairs)):
d[i] = {
'a': key[0],
'b': key[1],
'supp1': round(frequent_pairs[key][0],6),
'cond1': round(frequent_pairs[key][1],6),
'lift1': round(frequent_pairs[key][2],6),
'supp2': round(frequent_pairs[key][3],6),
'cond2': round(frequent_pairs[key][4],6),
'lift2': round(frequent_pairs[key][5],6),
}
df = pd.DataFrame.from_dict(d, "index")
df.sort_values(by=["b", "lift2"], inplace=True)
df.to_csv("./word_pair_2.csv", index=None)