rewrite to Keras function API

hw6/Model.py

@@ -0,0 +1,41 @@
+import numpy as np
+from keras.layers import Input, Embedding, Reshape, Dense, Dropout
+from keras.layers.merge import concatenate, dot
+from keras.models import Model
+
+def build_cf_model(n_users, n_movies, dim):
+ u_input = Input(shape=(1,))
+ u = Embedding(n_users, dim)(u_input)
+ u = Reshape((dim,))(u)
+
+ m_input = Input(shape=(1,))
+ m = Embedding(n_movies, dim)(m_input)
+ m = Reshape((dim,))(m)
+
+ out = dot([u, m], -1)
+
+ model = Model(inputs=[u_input, m_input], outputs=out)
+ return model
+
+
+def build_deep_model(n_users, n_movies, dim, dropout=0.1):
+ u_input = Input(shape=(1,))
+ u = Embedding(n_users, dim)(u_input)
+ u = Reshape((dim,))(u)
+
+ m_input = Input(shape=(1,))
+ m = Embedding(n_movies, dim)(m_input)
+ m = Reshape((dim,))(m)
+
+ out = concatenate([u, m])
+ out = Dropout(dropout)(out)
+ out = Dense(dim, activation='relu')(out)
+ out = Dropout(dropout)(out)
+ out = Dense(1, activation='linear')(out)
+
+ model = Model(inputs=[u_input, m_input], outputs=out)
+ return model
+
+def rate(model, user_id, item_id):
+ return model.predict([np.array([user_id]), np.array([item_id])])[0][0]
+

hw6/test.py

@@ -3,17 +3,18 @@
 import argparse
 import numpy as np
 import pandas as pd
-from CFModel import CFModel, DeepModel
+from Model import build_cf_model, build_deep_model, rate
+
 
 def parse_args():
  parser = argparse.ArgumentParser(description='HW6: Matrix Factorization')
- parser.add_argument('train', type=str)
+ parser.add_argument('data_dir', type=str)
  parser.add_argument('output', type=str)
  return parser.parse_args()
 
 
 def predict_rating(trained_model, userid, movieid):
- return trained_model.rate(userid - 1, movieid - 1)
+ return rate(trained_model, userid - 1, movieid - 1)
 
 
 def ensure_dir(file_path):
@@ -24,11 +25,6 @@ def ensure_dir(file_path):
 
 
 def main(args):
- ratings = pd.read_csv(args.train, usecols=['UserID', 'MovieID', 'Rating'])
- max_userid = ratings['UserID'].drop_duplicates().max()
- max_movieid = ratings['MovieID'].drop_duplicates().max()
- print('{} ratings loaded.'.format(ratings.shape[0]))
-
  users = pd.read_csv(USERS_CSV, sep='::', engine='python',
  usecols=['UserID', 'Gender', 'Age', 'Occupation', 'Zip-code'])
  print('{} description of {} users loaded'.format(len(users), max_userid))
@@ -40,7 +36,7 @@ def main(args):
  test_data = pd.read_csv(TEST_CSV, usecols=['UserID', 'MovieID'])
  print('{} testing data loaded.'.format(test_data.shape[0]))
 
- trained_model = DeepModel(max_userid, max_movieid, DIM)
+ trained_model = build_deep_model(max_userid, max_movieid, DIM)
  print('Loading model weights...')
  trained_model.load_weights(MODEL_WEIGHTS_FILE)
  print('Loading model done!!!')
@@ -56,12 +52,23 @@ def main(args):
 if __name__ == '__main__':
  args = parse_args()
 
- TEST_CSV = 'data/test.csv'
- USERS_CSV = 'data/users.csv'
- MOVIES_CSV = 'data/movies.csv'
+ MODEL_DIR = './model'
+ MAX_CSV = 'max.csv'
+ TEST_CSV = 'test.csv'
+ USERS_CSV = 'users.csv'
+ MOVIES_CSV = 'movies.csv'
  MODEL_WEIGHTS_FILE = 'weights.h5'
 
- DIM = 120
- TEST_USER = 3000
+ DATA_DIR = args.data_dir
+ TEST_CSV = os.path.join(DATA_DIR, TEST_CSV)
+ USERS_CSV = os.path.join(DATA_DIR, USERS_CSV)
+ MOVIES_CSV = os.path.join(DATA_DIR, MOVIES_CSV)
+
+ MODEL_WEIGHTS_FILE = os.path.join(MODEL_DIR, MODEL_WEIGHTS_FILE)
+ MAX_CSV = os.path.join(MODEL_DIR, MAX_CSV)
+ info = pd.read_csv(MAX_CSV)
+ DIM = list(info['dim'])[0]
+ max_userid = list(info['max_userid'])[0]
+ max_movieid = list(info['max_movieid'])[0]
 
  main(args)

hw6/train.py

@@ -4,12 +4,14 @@
 import numpy as np
 import pandas as pd
 from keras.callbacks import Callback, EarlyStopping, ModelCheckpoint
-from CFModel import CFModel, DeepModel
+from Model import build_cf_model, build_deep_model, rate
+
 
 def parse_args():
  parser = argparse.ArgumentParser(description='HW6: Matrix Factorization')
  parser.add_argument('train', type=str)
  parser.add_argument('test', type=str)
+ parser.add_argument('--dim', type=str, default=120)
  return parser.parse_args()
 
 
@@ -22,6 +24,13 @@ def main(args):
  ratings['Movie_emb_id'] = ratings['MovieID'] - 1
  print('{} ratings loaded.'.format(ratings.shape[0]))
 
+ maximum = {}
+ maximum['max_userid'] = [max_userid]
+ maximum['max_movieid'] = [max_movieid]
+ maximum['dim'] = [DIM]
+ pd.DataFrame(data=maximum).to_csv(MAX_FILE, index=False)
+ print('max info save to {}'.format(MAX_FILE))
+
  ratings = ratings.sample(frac=1)
  Users = ratings['User_emb_id'].values
  print('Users: {}, shape = {}'.format(Users, Users.shape))
@@ -30,7 +39,7 @@ def main(args):
  Ratings = ratings['Rating'].values
  print('Ratings: {}, shape = {}'.format(Ratings, Ratings.shape))
 
- model = DeepModel(max_userid, max_movieid, DIM)
+ model = build_deep_model(max_userid, max_movieid, DIM)
  model.compile(loss='mse', optimizer='adamax')
 
  callbacks = [EarlyStopping('val_loss', patience=2),
@@ -41,7 +50,13 @@ def main(args):
 if __name__ == '__main__':
  args = parse_args()
 
- DIM = 120
+ MODEL_DIR = './model'
+ DIM = args.dim
  MODEL_WEIGHTS_FILE = 'weights.h5'
+ MAX_FILE = 'max.csv'
 
+ if not os.path.exists(MODEL_DIR):
+ os.makedirs(MODEL_DIR)
+ MODEL_WEIGHTS_FILE = os.path.join(MODEL_DIR, MODEL_WEIGHTS_FILE)
+ MAX_FILE = os.path.join(MODEL_DIR, MAX_FILE)
  main(args)