w2vcnn

Author: sirent

W2VCNN.py

@@ -0,0 +1,204 @@
+import re
+
+import gensim
+import numpy as np
+import pandas as pd
+from keras.layers import Dense, Conv1D, MaxPooling1D, Flatten, Embedding, Dropout
+from keras.models import Sequential
+from keras.preprocessing.sequence import pad_sequences
+from keras.preprocessing.text import Tokenizer
+from nltk.corpus import stopwords
+from sklearn.model_selection import train_test_split
+
+
+###################################################
+def remove_stopwords(data):
+    stop_words = set(stopwords.words('indonesian'))
+    wordss = []
+    for sentence in data:
+        sentence = sentence.lower().split()
+        # print(sentence)
+        x = [word.strip() for word in sentence if word not in stop_words]
+        # print(x)
+        wordss.append(x)
+    return wordss
+
+
+def remove_punc_emoji(data):
+    # REMOVE EMOJI
+    preprocess = [re.sub(r'[^\x00-\x7F]+', '', sentence) for sentence in data]
+    # REMOVE ?
+    preprocess2 = [re.sub(r'\?+', '', sentence) for sentence in preprocess]
+    # REMOVE .
+    preprocess3 = [re.sub(r'\.+', ' ', sentence) for sentence in preprocess2]
+    # REMOVE HASHTAG
+    preprocess4 = [re.sub(r'(#[^\s]+)+', '', sentence) for sentence in preprocess3]
+    # REMOVE ' '
+    preprocess5 = [re.sub(r'', '', sentence) for sentence in preprocess4]
+    # REMOVE ,
+    preprocess6 = [re.sub(r',', ' ', sentence) for sentence in preprocess5]
+
+    return preprocess6
+
+
+###################################################
+
+# OPEN TRAIN DATA USING PANDAS
+train_data = pd.read_csv("Training_Jokowimundurlahbatch1.csv", sep=';', encoding='latin-1')
+print(train_data.head())
+print("Total rows: {0}".format(len(train_data)))
+print(list(train_data))
+
+# OPEN TEST DATA USING PANDAS
+test_data = pd.read_csv('Test_data.csv', sep=';', encoding='latin-1')
+print(test_data.head())
+print("Total rows: {0}".format(len(test_data)))
+print(list(test_data))
+
+# Comment in train data
+print("\nComments")
+print("-----------")
+Train_Comments = train_data.Comment
+print(Train_Comments)
+# print(row[0])
+
+# Comment in test data
+print("\nComments")
+print("-----------")
+Test_Comments = test_data.Comment
+print(Test_Comments)
+
+# Labels in train data
+print("\nLabels")
+print("-----------")
+# labels = ['Label']
+# train_labels = train_data[labels]
+
+# train_labels = train_data.Label
+
+labels = train_data.Label.unique()
+dic = {}
+for i, Label in enumerate(labels):
+    dic[Label] = i
+train_labels = train_data.Label.apply(lambda x: dic[x])
+print(train_labels)
+
+# PREPROCESSING TRAINING DATA
+preprocessed_train_comment = remove_punc_emoji(Train_Comments)
+print(preprocessed_train_comment)
+
+# PREPROCESSING TEST DATA
+preprocessed_test_comment = remove_punc_emoji(Test_Comments)
+print(preprocessed_test_comment)
+
+# LOWER AND SPLIT SENTENCE TO WORDS ALSO REMOVE STOPWORDS
+train_words = remove_stopwords(preprocessed_train_comment)  # TRAINING DATA
+test_words = remove_stopwords(preprocessed_test_comment)  # TEST DATA
+
+# INPUT TO WORD2VEC(SIZE = DIMENSION, MIN_COUNT = MINIMAL NUMBER OF WORDS APPEAR WILL BE LEARNED, ITER = ITERATION OF
+# SENTENCE WILL BE LEARNED)
+Embedding_dim = 300
+train_word_model = gensim.models.Word2Vec(train_words, size=Embedding_dim, min_count=1, iter=10, sg=1, window=3)
+
+# CHECK SIMILARITY WORD BANGSAT
+# print(train_word_model.wv.most_similar(positive='bangsat'))
+
+# EMBED WORD2VEC VOCABULARY TO MATRIX
+
+# MAX_NB_WORDS = 100000
+# nb_words = min(MAX_NB_WORDS, len(train_word_model.wv.vocab))+1
+
+embedding_matrix = np.zeros((len(train_word_model.wv.vocab) + 1, Embedding_dim))
+# embedding_matrix = np.zeros((nb_words, Embedding_dim))
+for i, vec in enumerate(train_word_model.wv.vectors):
+    embedding_matrix[i] = vec
+print(train_word_model.wv.vocab)
+
+
+# how many features should the tokenizer extract
+features = 500
+tokenizer = Tokenizer(num_words=features)
+
+# fit the tokenizer on our text
+tokenizer.fit_on_texts(train_words)
+
+# get all words that the tokenizer knows
+word_index = tokenizer.word_index
+
+# put the tokens in a matrix
+X = tokenizer.texts_to_sequences(train_words)
+X = pad_sequences(X)
+
+# prepare the labels
+y = pd.get_dummies(train_labels)
+
+# split in train and test
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, shuffle=False)
+
+########################################################################################################################
+model = Sequential()
+model.add(
+    Embedding(len(train_word_model.wv.vocab) + 1, Embedding_dim, input_length=X.shape[1], weights=[embedding_matrix],
+              trainable=False))
+
+model.add(Dropout(0.2))
+
+model.add(Conv1D(300,3,padding='valid',activation='relu',strides=2))
+model.add(Conv1D(150,3,padding='valid',activation='relu',strides=2))
+model.add(Conv1D(75,3,padding='valid',activation='relu',strides=2))
+
+# model.add(Conv1D(filters=16, kernel_size=3, activation='relu'))
+# model.add(Conv1D(filters=16, kernel_size=4, activation='relu'))
+# model.add(Conv1D(filters=16, kernel_size=5, activation='relu'))
+# model.add(Conv1D(filters=16, kernel_size=8, activation='relu'))
+
+# model.add(Conv1D(filters=32, kernel_size=3, activation='relu'))
+# model.add(Conv1D(filters=32, kernel_size=4, activation='relu'))
+# model.add(Conv1D(filters=32, kernel_size=5, activation='relu'))
+# model.add(Conv1D(filters=32, kernel_size=8, activation='relu'))
+
+# model.add(Conv2D(filters=100, kernel_size=8, activation='relu', kernel_regularizer=regularizers.l2(0.01)))
+# model.add(Conv2D(filters=100, kernel_size=4, activation='relu', kernel_regularizer=regularizers.l2(0.01)))
+# model.add(Conv2D(filters=100, kernel_size=5, activation='relu', kernel_regularizer=regularizers.l2(0.01)))
+
+model.add(MaxPooling1D(pool_size=2))
+# model.add(MaxPooling2D(pool_size=2))
+
+model.add(Flatten())
+model.add(Dropout(0.2))
+# model.add(Dense(10, activation='relu'))
+
+model.add(Dense(150, activation='sigmoid'))
+
+model.add(Dropout(0.2))
+# model.add(Dense(1, activation='sigmoid'))
+
+model.add(Dense(3, activation='sigmoid'))
+
+# model.add(Dense(y.shape[1], activation="softmax"))
+model.add(Dense(y.shape[1], activation="sigmoid"))
+
+print(model.summary())
+# model.compile(optimizer="adam", loss="categorical_crossentropy", metrics=['acc'])
+model.compile(optimizer="rmsprop", loss="categorical_crossentropy", metrics=['acc'])
+# model.compile(loss='mean_squared_error', optimizer='sgd', metrics=['mae', 'acc'])
+#############################################
+
+# EVALUATION TEST
+batch = 128
+epo = 2
+model.fit(X_train, y_train, validation_data=(X_test, y_test), batch_size=batch, epochs=epo)
+
+evaluate = model.evaluate(X_test, y_test, verbose=0)
+print('\nTest Loss:', evaluate[0])
+print('Test Accuracy:', evaluate[1])
+###############################################
+
+# REAL DATA TEST
+sequences_test = tokenizer.texts_to_sequences(test_data)
+X_real_test = pad_sequences(sequences_test, maxlen=X_train.shape[1])
+# X_real_test = pad_sequences(sequences_test, maxlen=MAX_SEQUENCE_LENGTH)
+y_pred = model.predict(X_real_test)
+to_submit = pd.DataFrame(index=test_data.Id, data={'Label': y_pred[:, dic['Label']]})
+to_submit.to_csv('submit.csv')
+##############################################