ts/data_preprocess.py at master · linhthi/ts · 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
import random
import pickle
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split

random.seed(1234)


class dataPreprocess(object):

    def __init__(self, filepath):
        self.filepath = filepath

    def preprocess(self):
        '''
        Description: Function to preprocess the data and store as '.pkl' files
        INPUT: Path to the dataset folder (keep the dataset as 'rating.csv' for ratings data and 'trustnetwork.csv' for social trust data
        OUTPUT: This function doesn't return anything but stores the data is '.pkl' files at the same folder.
        '''

        ratingsData = pd.read_csv(self.filepath + '/rating.csv').values
        trustData = pd.read_csv(self.filepath + '/trustnetwork.csv').values

        ratingsList = []
        trustList = []

        users = set()
        items = set()

        for row in ratingsData:
            userId = row[0]
            itemId = row[1]
            rating = row[3]
            if userId not in users:
                users.add(userId)
            if itemId not in items:
                items.add(itemId)
            ratingsList.append([userId, itemId, rating])

        userCount = len(users)
        itemCount = len(items)

        for row in trustData:
            user1 = row[0]
            user2 = row[1]
            trust = 1
            trustList.append([user1, user2, trust])

        newDF = pd.DataFrame(ratingsList, columns=['userId', 'itemId', 'rating'])
        X = np.array([newDF['userId'], newDF['itemId']]).T
        y = np.array([newDF['rating']]).T

        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0, stratify=y)
        train = pd.DataFrame(X_train, columns=['userId', 'itemId'])
        train['rating'] = pd.DataFrame(y_train)
        test = pd.DataFrame(X_test, columns=['userId', 'itemId'])
        test['rating'] = pd.DataFrame(y_test)

        trainUsers = []
        trainItems = []
        trainRatings = []
        for index in range(len(train)):
            trainUsers.append(train['userId'][index])
            trainItems.append(train['itemId'][index])
            trainRatings.append(train['rating'][index])

        testUsers = []
        testItems = []
        testRatings = []
        for index in range(len(test)):
            testUsers.append(test['userId'][index])
            testItems.append(test['itemId'][index])
            testRatings.append(test['rating'][index])

        userItemDict = {}
        for index in range(len(train)):
            if train['userId'][index] not in userItemDict:
                userItemDict[train['userId'][index]] = [train['itemId'][index]]
            else:
                userItemDict[train['userId'][index]].append(train['itemId'][index])

        userRatings = {}
        for index in range(len(train)):
            if train['userId'][index] not in userRatings:
                userRatings[train['userId'][index]] = [train['rating'][index]]
            else:
                userRatings[train['userId'][index]].append(train['rating'][index])

        itemUserDict = {}
        for index in range(len(train)):
            if train['itemId'][index] not in itemUserDict:
                itemUserDict[train['itemId'][index]] = [train['userId'][index]]
            else:
                itemUserDict[train['itemId'][index]].append(train['userId'][index])

        itemRatings = {}
        for index in range(len(train)):
            if train['itemId'][index] not in itemRatings:
                itemRatings[train['itemId'][index]] = [train['rating'][index]]
            else:
                itemRatings[train['itemId'][index]].append(train['rating'][index])

        trust = pd.DataFrame(trustList, columns=['userId', 'friendID', 'trust'])

        userUserDict = {}
        for index in range(len(trust)):
            if trust['userId'][index] not in userUserDict:
                userUserDict[trust['userId'][index]] = {trust['friendID'][index]}
            else:
                userUserDict[trust['userId'][index]].add(trust['friendID'][index])
            if trust['friendID'][index] not in userUserDict:
                userUserDict[trust['friendID'][index]] = {trust['userId'][index]}
            else:
                userUserDict[trust['friendID'][index]].add(trust['userId'][index])

        ratings = []
        for i in userRatings.keys():
            ratings.append(userRatings[i])
        r = [i for row in ratings for i in row]
        r = list(set(r))
        ratingsL = {}
        for i in range(1, len(r) + 1):
            if i not in ratingsL.keys():
                ratingsL[i] = r[i - 1]
            else:
                continue

        user = []
        friend = []
        trusts = []
        for index in range(len(trust)):
            user.append(trust['userId'][index])
            friend.append(trust['friendID'][index])
            trusts.append(trust['trust'][index])
        user = list(set(user))
        friend = list(set(friend))

        with open(self.filepath + '/dataset.pickle', 'wb') as files:
            pickle.dump((userItemDict, userRatings, itemUserDict, itemRatings,
                        trainUsers, trainItems, trainRatings,
                        testUsers, testItems, testRatings,
                        userUserDict, ratingsL, user, friend, trust), files, pickle.HIGHEST_PROTOCOL)