add rmse metrics & fix some bugs

hw6/Model.py

@@ -31,7 +31,7 @@ def build_deep_model(n_users, n_movies, dim, dropout=0.1):
  out = Dropout(dropout)(out)
  out = Dense(dim, activation='relu')(out)
  out = Dropout(dropout)(out)
- out = Dense(1, activation='linear')(out)
+ out = Dense(1, activation='relu')(out)
 
  model = Model(inputs=[u_input, m_input], outputs=out)
  return model

hw6/test.py

@@ -36,7 +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 = build_deep_model(max_userid, max_movieid, DIM)
+ trained_model = build_cf_model(max_userid, max_movieid, DIM)
  print('Loading model weights...')
  trained_model.load_weights(MODEL_WEIGHTS_FILE)
  print('Loading model done!!!')

hw6/train.py

@@ -3,6 +3,7 @@
 import argparse
 import numpy as np
 import pandas as pd
+from keras import backend as K
 from keras.callbacks import Callback, EarlyStopping, ModelCheckpoint
 from Model import build_cf_model, build_deep_model, rate
 
@@ -11,10 +12,14 @@ 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)
+ parser.add_argument('--dim', type=int, default=120)
  return parser.parse_args()
 
 
+def rmse(y_true, y_pred):
+ y_pred = K.clip(y_pred, 1., 5.)
+ return K.sqrt(K.mean(K.square((y_true - y_pred))))
+
 def main(args):
  ratings = pd.read_csv(args.train,
  usecols=['UserID', 'MovieID', 'Rating'])
@@ -39,12 +44,12 @@ def main(args):
  Ratings = ratings['Rating'].values
  print('Ratings: {}, shape = {}'.format(Ratings, Ratings.shape))
 
- model = build_deep_model(max_userid, max_movieid, DIM)
- model.compile(loss='mse', optimizer='adamax')
+ model = build_cf_model(max_userid, max_movieid, DIM)
+ model.compile(loss='mse', optimizer='adamax', metrics=[rmse])
 
- callbacks = [EarlyStopping('val_loss', patience=2),
+ callbacks = [EarlyStopping('val_rmse', patience=2),
  ModelCheckpoint(MODEL_WEIGHTS_FILE, save_best_only=True)]
- history = model.fit([Users, Movies], Ratings, epochs=100, validation_split=.1, verbose=1, callbacks=callbacks)
+ history = model.fit([Users, Movies], Ratings, epochs=100, batch_size=256, validation_split=.1, verbose=1, callbacks=callbacks)
 
 
 if __name__ == '__main__':