서버 연동 후 코드 수정

Author: k2river

AI/ALS.py

@@ -1,8 +1,9 @@
 import pandas as pd
 import numpy as np
 
-def get_ratings(path, users_file_name):
-    df_users = pd.read_csv(path + users_file_name, encoding='cp949')
+def get_ratings(path, users_file_name, books_file_name):
+    df_users = pd.read_csv(path + users_file_name, encoding='UTF8')
+    df_books = pd.read_csv(path + books_file_name, encoding='UTF8')
     df_users_books = pd.DataFrame(df_users, columns=['user_id', 'like'])
     sr_users = []
     sr_books = []
@@ -14,19 +15,28 @@ def get_ratings(path, users_file_name):
         i = i.split(', ')
         list_like.append(i)
 
-    for user_id in df_users_books['user_id']:
-        for book_id in list_like[user_id]:
-            sr_users.append(user_id)
-            sr_books.append(book_id)
+    for user_idx in df_users_books['user_id']:
+        for book_idx in list_like[user_idx]:
+            if book_idx == '':
+                break
+            sr_users.append(user_idx)
+            sr_books.append(book_idx)
             sr_ratings.append(1)
 
+    for book_idx in range(len(df_books)):
+        if sr_users[0]:
+            sr_users.append(sr_users[0])
+        if sr_books[0]:
+            sr_books.append(book_idx)
+        if sr_ratings[0]:
+            sr_ratings.append(0)
     R = pd.DataFrame({
-        'user_id': sr_users,
-        'book_id': sr_books,
+        'user_idx': sr_users,
+        'book_idx': sr_books,
         'ratings': sr_ratings
     })
 
-    R = R.pivot_table('ratings', index='user_id', columns='book_id').fillna(0)
+    R = R.pivot_table('ratings', index='user_idx', columns='book_idx').fillna(0)
     R.rename(columns= lambda x: int(x), inplace=True)
     R = R.sort_index(axis=1)
     return R
@@ -76,7 +86,7 @@ def predict(file_path, users_file_name, pred_score_file_name):
     regularization_list = []
     total_losses = []
 
-    for i in range(15):
+    for i in range(6):
         if i != 0:
             optimize_user(X, Y, C, P, nu, nf, r_lambda)
             optimize_item(X, Y, C, P, ni, nf, r_lambda)
@@ -100,11 +110,3 @@ def predict(file_path, users_file_name, pred_score_file_name):
 
     df_predict = pd.DataFrame(predict, columns=range(len(R[0]))).fillna(0)  # user-item = 1400 x 1125
     df_predict.to_csv(file_path + pred_score_file_name)
-
-'''
-file_path = '/var/www/python_flask/main/models/'
-users_file_name = "API_test_users.csv"
-books_file_name = "API_test_books.csv"
-pred_file_name = "rec_pred_score_1.csv"
-
-predict(file_path, users_file_name, pred_file_name)'''

AI/data_update.py

@@ -0,0 +1,47 @@
+import os
+import sys
+sys.path.append(os.path.dirname(os.path.abspath(os.path.dirname(__file__))))
+sys.path.append(os.path.dirname(os.path.abspath(os.path.dirname(os.path.abspath(os.path.dirname(__file__))))))
+from main.models import database
+import pandas as pd
+import numpy as np
+
+def update():
+    '''
+    books = database.Book.objects()
+    isbn = []
+    for b in books:
+        isbn.append(b.isbn)
+    '''
+    file_path = '/var/www/python_flask/main/recommendation/'
+    books_file_name = 'API_test_books.csv'
+    df_books = pd.read_csv(file_path + books_file_name)
+    isbn = list(np.array(df_books['isbn']).tolist())
+
+    dummy_file_name = 'dummy_users.csv'
+    df_dummy_user = pd.read_csv(file_path + dummy_file_name)
+
+    users = database.User.objects()
+    name = []
+    email = []
+    password = []
+    user_id = []
+    like = []
+    rank = []
+    unit = []
+    for u in users:
+        t_list = []
+        for l in u.like:
+            t = l.rstrip('/')
+            try:
+                t = isbn.index(t)
+            except:
+                continue
+            t_list.append(t)
+        like.append(t_list)
+        name.append(u.name)
+        email.append(u.email)
+        password.append(u.password)
+        user_id.append(u.user_id)
+        rank.append(u.rank)
+        unit.append(u.unit)

AI/dummy_generator.py

@@ -0,0 +1,80 @@
+import pandas as pd
+import random
+
+file_path = '/var/www/python_flask/main/recommendation/'  # 서버 폴더경로 맞춰서 다시 설정
+save_path = 'recommend_list/'
+users_file_name = "dummy_users.csv"
+books_file_name = "API_test_books.csv"
+categories_file_name = "rec_category_2.csv"
+
+df_books = pd.read_csv(file_path + books_file_name, encoding='cp949')
+df_category = pd.read_csv(file_path + categories_file_name, encoding='cp949')
+df_category = df_category[['category', 'count']].dropna()
+
+def make_like_list(category):
+    cat_idx = category
+    cat_count = df_category.iloc[cat_idx]['count']
+    book_list = list(df_books.index[df_books['categoryName'] == df_category.iloc[cat_idx]['category']])
+    pick_count = random.randint(1, min(10, cat_count))
+    like_list = []
+    for i in range(pick_count):
+        pick_idx = random.randint(0, cat_count-1)
+        cnt = 0
+        while book_list[pick_idx] in like_list and cnt < 10:
+            pick_idx = random.randint(0, cat_count - 1)
+        if cnt < 10:
+            like_list.append(book_list[pick_idx])
+    return like_list
+
+def make_like_category():
+    cat_len = len(df_category)
+    cat_num = random.randint(1, min(10, cat_len))
+    cat_list = []
+    like_list = []
+    for i in range(cat_num):
+        pick_cat = random.randint(0, cat_len-1)
+        cnt = 0
+        while pick_cat in cat_list and cnt < 10:
+            pick_cat = random.randint(0, cat_len - 1)
+        if cnt < 10:
+            cat_list.append(pick_cat)
+
+    for category in cat_list:
+        like_list += make_like_list(category)
+
+    return like_list
+
+like_list = []
+for i in range(50):
+    like_list.append(make_like_category())
+
+def make_dataframe(like_list):
+    n = len(like_list)
+    list_unit = []
+    list_rank = []
+    list_user_id = []
+
+    int_rank = ['이등병', '일병', '상병', '병장']
+    for i in range(n):
+        list_unit.append("53사단" if random.randint(0, 1) else "31사단")
+        list_rank.append(int_rank[random.randint(0, 3)])
+
+    for i in range(n):
+        list_user_id.append('userid_'+str(i))
+
+    to_df = {
+        'name': range(n),
+        'email': range(n),
+        'password': range(n),
+        'user_id': list_user_id,
+        'like':like_list,
+        'rank': list_rank,
+        'unit': list_unit
+    }
+
+    df = pd.DataFrame(to_df)
+    df = df.fillna(0)
+    return df
+
+df = make_dataframe(like_list)
+df.to_csv(file_path + users_file_name)