forked from 827983519/recommend_algorithm
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathGA_item.py
More file actions
192 lines (162 loc) · 7.4 KB
/
GA_item.py
File metadata and controls
192 lines (162 loc) · 7.4 KB
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
from sklearn.decomposition import PCA
import numpy as np
import pandas as pd
from sklearn.metrics import mean_squared_error
from sklearn.cluster import KMeans
from sklearn.model_selection import train_test_split
from sklearn.metrics import pairwise_distances
def fresh_property(population_current):
sum = 0
for i in range(Population_size):
population_current[i][1]['fitness'] = objective_function(population_current[i])
sum += population_current[i][1]['fitness']
population_current[0][1]['rate_fit'] = population_current[0][1]['fitness'] / sum
population_current[0][1]['cumm_fit'] = population_current[0][1]['rate_fit']
for i in range(Population_size):
population_current[i][1]['rate_fit'] = population_current[i][1]['fitness'] / sum
population_current[i][1]['cumm_fit'] = population_current[i][1]['rate_fit'] + population_current[i-1][1]['cumm_fit']
def objective_function(individual):
distance = 0
min_distance = 100000000
for row in range(len(pca_data)):
for i in range(Cluster_number):
euclidean_distance = np.abs(pca_data.iloc[row].values - individual[0][i]).sum()
if euclidean_distance < min_distance:
min_distance = euclidean_distance
distance += min_distance
return distance
def select(population_current,population_next):
for i in range(Population_size):
rand = np.random.rand(1)
if rand <= population_current[0][1]['cumm_fit']:
population_next[i] = population_current[0]
else:
for j in range(Population_size):
if population_current[j][1]['cumm_fit'] <= rand and population_current[j+1][1]['cumm_fit'] >= rand:
population_next[i] = population_current[j+1]
break
def crossover(population_next):
for i in range(Population_size):
rand = np.random.rand(1)
if rand <= Probability_crossover:
rand_cluster = np.random.randint(Cluster_number)
p1_num = np.random.randint(Population_size)
p2_num = np.random.randint(Population_size)
p1 = population_next[p1_num]
p2 = population_next[p2_num]
c1 = p1
c2 = p2
c1[0] = np.vstack([p1[0][:rand_cluster,:],p2[0][rand_cluster:,:]])
c2[0] = np.vstack([p2[0][:rand_cluster,:],p1[0][rand_cluster:,:]])
test_c = [[],[]]
test_c[0].extend([objective_function(c1),objective_function(c2),objective_function(p1),objective_function(p2)])
test_c[1].extend([c1,c2,p1,p2])
population_next[p1_num] = test_c[1][test_c[0].index(min(test_c[0]))]
test_c[1] = test_c[1][:test_c[0].index(min(test_c[0]))] + test_c[1][test_c[0].index(min(test_c[0]))+1:]
test_c[0].remove(min(test_c[0]))
population_next[p2_num] = test_c[1][test_c[0].index(min(test_c[0]))]
def mutation(population_next):
for i in range(Population_size):
rand = np.random.rand(1)
if rand <= Probability_mutation:
mutation_array = np.ones([Cluster_number,Dimension_number])
for k in range(Cluster_number):
rand_pick = np.random.randint(Population_size)
mutation_array[k] = population_next[rand_pick][0][k]
if objective_function([mutation_array]) < objective_function(population_next[i]):
population_next[i][0] = mutation_array
def user_predict(train_data,label,euc,test_data,k,user_mean,movie_mean):
prediction_set = []
new = 0
last = 0
for i in range(len(test_data)):
user = int(test_data.iloc[i].userId)
movie = int(test_data.iloc[i].movieId)
new = user
if movie not in train_data.index:
prediction_set.append(user_mean[user])
continue
if new != last:
cluster_label = label.loc[movie].label
mean = movie_mean[movie]
k_similar_index = []
if len(label[label['label'] == cluster_label]) < k:
k_similar_index.extend(euc[movie][label[label['label'] == cluster_label].index].index)
else:
k_similar_index.extend(euc[movie][label[label['label'] == cluster_label].index].sort_values(ascending=True)[:k].index)
add_up = 0
add_down = 0
for similar_index in k_similar_index:
if train_data[user][similar_index] != 0:
similar_mean = movie_mean[similar_index]
add_up = add_up + euc[movie][similar_index] * (train_data[user][similar_index] - similar_mean)
add_down = add_down + abs(euc[movie][similar_index])
if add_down == 0:
prediction = mean
else:
prediction = mean+add_up/add_down
if(prediction > 5):
prediction = 5
if(prediction < 0):
predition = 0
prediction_set.append(prediction)
last = new
return prediction_set
error_set = []
#ga_file = open('ga','w')
rating = pd.read_csv('ratings.dat',sep='::',header=None,names=['userId','movieId','rating','timestamp'])
train,test = train_test_split(rating,test_size = 0.2,random_state=0)
test = test.sort_index()
item_based = train.pivot_table(index='movieId', columns='userId', values='rating')
user_mean = item_based.mean(axis=0)
movie_mean = item_based.mean(axis=1)
item_based = item_based.fillna(0)
pca = PCA(n_components=500)
pca_data = pd.DataFrame(pca.fit_transform(item_based),index=item_based.index)
min_max = []
min_max.append(pca_data.max())
min_max.append(pca_data.min())
cluster_num_set = [i for i in range(3,16,2)]
for cluster_num in range(3,15,1):
print('cluster_num',cluster_num)
Population_size = 50
Cluster_number = cluster_num
Dimension_number = 500
iteration_num = 140
Probability_crossover = 0.5
Probability_mutation = 0.0001;
population_current = []
population_next = []
for i in range(Population_size):
gene_array = np.array([])
for j in range(Dimension_number):
gene = np.random.uniform(min_max[0][j],min_max[1][j],(Cluster_number,1))
if len(gene_array) == 0:
gene_array = gene
else:
gene_array = np.hstack([gene_array,gene])
population_current.append([gene_array,{'rate_fit':0,'cumm_fit':0,'fitness':0}])
population_next = population_current[:]
fresh_property(population_current)
for i in range(iteration_num):
print('iteration',i)
select(population_current,population_next)
crossover(population_next)
mutation(population_next)
fresh_property(population_next)
population_current = population_next[:]
kmeans = KMeans(n_clusters=cluster_num,init=population_next[0][0])
kmeans.fit(pca_data)
label = pd.DataFrame(kmeans.labels_,index = item_based.index,columns=['label'])
route = str(cluster_num)+'.csv'
label.to_csv(route)
'''
euc = pd.DataFrame(pairwise_distances(item_based,metric='euclidean'),index=item_based.index,columns=item_based.index)
error = np.sqrt(mean_squared_error(test.rating,user_predict(item_based,label,euc,test,100,user_mean,movie_mean)))
ga_file.write('cluseter_num ')
ga_file.write(str(cluster_num))
ga_file.write(': ')
ga_file.write(str(error))
ga_file.write('\n')
ga_file.close()
'''