get-started-with-sentiment-analysis-tf-keras.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.10","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Get Started With Sentiment Analysis Using TensorFlow Keras\n\nReferences:\n\nhttps://www.tensorflow.org/datasets/catalog/imdb_reviews\n\nhttps://www.tensorflow.org/text/tutorials/text_classification_rnn\n","metadata":{}},{"cell_type":"code","source":"import numpy as np\nimport tensorflow as tf\nimport tensorflow_datasets as tfds\nfrom tensorflow import keras\nfrom tensorflow.keras.layers import Dense, Dropout, Bidirectional, LSTM\nimport matplotlib.pyplot as plt","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2021-05-26T00:39:59.956795Z","iopub.execute_input":"2021-05-26T00:39:59.957348Z","iopub.status.idle":"2021-05-26T00:40:06.792029Z","shell.execute_reply.started":"2021-05-26T00:39:59.957236Z","shell.execute_reply":"2021-05-26T00:40:06.790876Z"},"trusted":true},"execution_count":1,"outputs":[]},{"cell_type":"markdown","source":"# IMDB Dataset","metadata":{}},{"cell_type":"code","source":"data, meta = tfds.load('imdb_reviews/subwords8k',\n                      with_info = True,\n                      as_supervised = True)","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:40:06.793718Z","iopub.execute_input":"2021-05-26T00:40:06.794053Z","iopub.status.idle":"2021-05-26T00:43:13.009835Z","shell.execute_reply.started":"2021-05-26T00:40:06.794018Z","shell.execute_reply":"2021-05-26T00:43:13.008526Z"},"trusted":true},"execution_count":2,"outputs":[{"name":"stdout","text":"\u001b[1mDownloading and preparing dataset imdb_reviews/subwords8k/1.0.0 (download: 80.23 MiB, generated: Unknown size, total: 80.23 MiB) to /root/tensorflow_datasets/imdb_reviews/subwords8k/1.0.0...\u001b[0m\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"Dl Completed...: 0 url [00:00, ? url/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"557209eac8704a868eb25aa07e356f98"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Dl Size...: 0 MiB [00:00, ? MiB/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"3e539c94f1c74f34bcc9d1da09051f14"}},"metadata":{}},{"name":"stdout","text":"\n\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"0 examples [00:00, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":""}},"metadata":{}},{"name":"stdout","text":"Shuffling and writing examples to /root/tensorflow_datasets/imdb_reviews/subwords8k/1.0.0.incompleteL9QVVU/imdb_reviews-train.tfrecord\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/25000 [00:00<?, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"6c139c71969241f5a72e2ae29e969f3c"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"0 examples [00:00, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":""}},"metadata":{}},{"name":"stdout","text":"Shuffling and writing examples to /root/tensorflow_datasets/imdb_reviews/subwords8k/1.0.0.incompleteL9QVVU/imdb_reviews-test.tfrecord\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/25000 [00:00<?, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"3e528f96c25e4a84ad1900a547108ec3"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"0 examples [00:00, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":""}},"metadata":{}},{"name":"stdout","text":"Shuffling and writing examples to /root/tensorflow_datasets/imdb_reviews/subwords8k/1.0.0.incompleteL9QVVU/imdb_reviews-unsupervised.tfrecord\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/50000 [00:00<?, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"4a3b039411d24b62a0b8d26377aa527d"}},"metadata":{}},{"name":"stdout","text":"\u001b[1mDataset imdb_reviews downloaded and prepared to /root/tensorflow_datasets/imdb_reviews/subwords8k/1.0.0. Subsequent calls will reuse this data.\u001b[0m\n","output_type":"stream"}]},{"cell_type":"code","source":"data.keys()","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.011936Z","iopub.execute_input":"2021-05-26T00:43:13.012262Z","iopub.status.idle":"2021-05-26T00:43:13.020526Z","shell.execute_reply.started":"2021-05-26T00:43:13.012228Z","shell.execute_reply":"2021-05-26T00:43:13.019454Z"},"trusted":true},"execution_count":3,"outputs":[{"execution_count":3,"output_type":"execute_result","data":{"text/plain":"dict_keys(['test', 'train', 'unsupervised'])"},"metadata":{}}]},{"cell_type":"code","source":"train = data['train']\ntest = data['test']\ntrain, test","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.022848Z","iopub.execute_input":"2021-05-26T00:43:13.023247Z","iopub.status.idle":"2021-05-26T00:43:13.039196Z","shell.execute_reply.started":"2021-05-26T00:43:13.023211Z","shell.execute_reply":"2021-05-26T00:43:13.038243Z"},"trusted":true},"execution_count":4,"outputs":[{"execution_count":4,"output_type":"execute_result","data":{"text/plain":"(<PrefetchDataset shapes: ((None,), ()), types: (tf.int64, tf.int64)>,\n <PrefetchDataset shapes: ((None,), ()), types: (tf.int64, tf.int64)>)"},"metadata":{}}]},{"cell_type":"code","source":"# explore the features in metadata\nmeta.features","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.040832Z","iopub.execute_input":"2021-05-26T00:43:13.041318Z","iopub.status.idle":"2021-05-26T00:43:13.055356Z","shell.execute_reply.started":"2021-05-26T00:43:13.041266Z","shell.execute_reply":"2021-05-26T00:43:13.054548Z"},"trusted":true},"execution_count":5,"outputs":[{"execution_count":5,"output_type":"execute_result","data":{"text/plain":"FeaturesDict({\n    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=2),\n    'text': Text(shape=(None,), dtype=tf.int64, encoder=<SubwordTextEncoder vocab_size=8185>),\n})"},"metadata":{}}]},{"cell_type":"code","source":"# extract the encoder\nencoder = meta.features['text'].encoder","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.056661Z","iopub.execute_input":"2021-05-26T00:43:13.057142Z","iopub.status.idle":"2021-05-26T00:43:13.069121Z","shell.execute_reply.started":"2021-05-26T00:43:13.057078Z","shell.execute_reply":"2021-05-26T00:43:13.067744Z"},"trusted":true},"execution_count":6,"outputs":[]},{"cell_type":"code","source":"encoder.vocab_size","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.070530Z","iopub.execute_input":"2021-05-26T00:43:13.070957Z","iopub.status.idle":"2021-05-26T00:43:13.084264Z","shell.execute_reply.started":"2021-05-26T00:43:13.070922Z","shell.execute_reply":"2021-05-26T00:43:13.083412Z"},"trusted":true},"execution_count":7,"outputs":[{"execution_count":7,"output_type":"execute_result","data":{"text/plain":"8185"},"metadata":{}}]},{"cell_type":"code","source":"print(encoder.subwords[:100])","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.086970Z","iopub.execute_input":"2021-05-26T00:43:13.087558Z","iopub.status.idle":"2021-05-26T00:43:13.099990Z","shell.execute_reply.started":"2021-05-26T00:43:13.087512Z","shell.execute_reply":"2021-05-26T00:43:13.098818Z"},"trusted":true},"execution_count":8,"outputs":[{"name":"stdout","text":"['the_', ', ', '. ', 'a_', 'and_', 'of_', 'to_', 's_', 'is_', 'br', 'in_', 'I_', 'that_', 'this_', 'it_', ' /><', ' />', 'was_', 'The_', 'as_', 't_', 'with_', 'for_', '.<', 'on_', 'but_', 'movie_', ' (', 'are_', 'his_', 'have_', 'film_', 'not_', 'ing_', 'be_', 'ed_', 'you_', ' \"', 'it', 'd_', 'an_', 'he_', 'by_', 'at_', 'one_', 'who_', 'y_', 'from_', 'e_', 'or_', 'all_', 'like_', 'they_', '\" ', 'so_', 'just_', 'has_', ') ', 'her_', 'about_', 'out_', 'This_', 'some_', 'ly_', 'movie', 'film', 'very_', 'more_', 'It_', 'would_', 'what_', 'when_', 'which_', 'good_', 'if_', 'up_', 'only_', 'even_', 'their_', 'had_', 'really_', 'my_', 'can_', 'no_', 'were_', 'see_', 'she_', '? ', 'than_', '! ', 'there_', 'get_', 'been_', 'into_', ' - ', 'will_', 'much_', 'story_', 'because_', 'ing']\n","output_type":"stream"}]},{"cell_type":"code","source":"example = 'Analytics India Magazine !'\nenc = encoder.encode(example)\nenc","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.101751Z","iopub.execute_input":"2021-05-26T00:43:13.102323Z","iopub.status.idle":"2021-05-26T00:43:13.115257Z","shell.execute_reply.started":"2021-05-26T00:43:13.102288Z","shell.execute_reply":"2021-05-26T00:43:13.114399Z"},"trusted":true},"execution_count":9,"outputs":[{"execution_count":9,"output_type":"execute_result","data":{"text/plain":"[1291, 375, 8050, 1540, 8, 6552, 4, 2853, 7908, 442, 5219]"},"metadata":{}}]},{"cell_type":"code","source":"for integer in enc:\n    text = encoder.decode([integer])\n    print('%4d : %s'%(integer, text))","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.116660Z","iopub.execute_input":"2021-05-26T00:43:13.117246Z","iopub.status.idle":"2021-05-26T00:43:13.131676Z","shell.execute_reply.started":"2021-05-26T00:43:13.117212Z","shell.execute_reply":"2021-05-26T00:43:13.130363Z"},"trusted":true},"execution_count":10,"outputs":[{"name":"stdout","text":"1291 : An\n 375 : al\n8050 : y\n1540 : tic\n   8 : s \n6552 : Indi\n   4 : a \n2853 : Mag\n7908 : azi\n 442 : ne\n5219 :  !\n","output_type":"stream"}]},{"cell_type":"code","source":"it = iter(train)","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.133447Z","iopub.execute_input":"2021-05-26T00:43:13.134133Z","iopub.status.idle":"2021-05-26T00:43:13.246380Z","shell.execute_reply.started":"2021-05-26T00:43:13.134090Z","shell.execute_reply":"2021-05-26T00:43:13.245500Z"},"trusted":true},"execution_count":11,"outputs":[]},{"cell_type":"code","source":"next(it)[0].numpy().shape, next(it)[1].numpy().shape","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.247752Z","iopub.execute_input":"2021-05-26T00:43:13.248263Z","iopub.status.idle":"2021-05-26T00:43:13.280044Z","shell.execute_reply.started":"2021-05-26T00:43:13.248215Z","shell.execute_reply":"2021-05-26T00:43:13.278949Z"},"trusted":true},"execution_count":12,"outputs":[{"execution_count":12,"output_type":"execute_result","data":{"text/plain":"((163,), ())"},"metadata":{}}]},{"cell_type":"code","source":"BUFFER_SIZE = 10000\nBATCH_SIZE = 64\nAUTOTUNE = tf.data.AUTOTUNE\n\ntrain_data = train.shuffle(BUFFER_SIZE)\ntrain_data = train_data.padded_batch(BATCH_SIZE, padded_shapes=([None],[]))\ntrain_data = train_data.prefetch(AUTOTUNE)\n\ntest_data = test.padded_batch(BATCH_SIZE, padded_shapes=([None],[]))","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.281483Z","iopub.execute_input":"2021-05-26T00:43:13.281796Z","iopub.status.idle":"2021-05-26T00:43:13.294522Z","shell.execute_reply.started":"2021-05-26T00:43:13.281764Z","shell.execute_reply":"2021-05-26T00:43:13.293448Z"},"trusted":true},"execution_count":13,"outputs":[]},{"cell_type":"code","source":"embed_layer = keras.layers.Embedding(encoder.vocab_size, 64)","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.296343Z","iopub.execute_input":"2021-05-26T00:43:13.296791Z","iopub.status.idle":"2021-05-26T00:43:13.309714Z","shell.execute_reply.started":"2021-05-26T00:43:13.296745Z","shell.execute_reply":"2021-05-26T00:43:13.308528Z"},"trusted":true},"execution_count":14,"outputs":[]},{"cell_type":"markdown","source":"# Build Model","metadata":{}},{"cell_type":"code","source":"model = keras.Sequential([\n    # embedding layer\n    embed_layer,\n    # bidirectional LSTM layers\n    Bidirectional(LSTM(64, \n                       dropout=0.5, \n                       recurrent_dropout=0.5, \n                       return_sequences=True)),\n    Bidirectional(LSTM(32, \n                       dropout=0.5, \n                       recurrent_dropout=0.5, \n                       return_sequences=True)),\n    Bidirectional(LSTM(16, \n                       dropout=0.5, \n                       recurrent_dropout=0.5)),\n    # Classification head\n    Dense(64, activation='relu', kernel_regularizer='l2'),\n    Dropout(0.5),\n    Dense(1, activation='sigmoid')    \n])","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:43:13.311212Z","iopub.execute_input":"2021-05-26T00:43:13.312012Z","iopub.status.idle":"2021-05-26T00:43:14.643839Z","shell.execute_reply.started":"2021-05-26T00:43:13.311977Z","shell.execute_reply":"2021-05-26T00:43:14.642775Z"},"trusted":true},"execution_count":15,"outputs":[]},{"cell_type":"code","source":"model.summary()","metadata":{"execution":{"iopub.status.busy":"2021-05-25T15:44:48.751754Z","iopub.execute_input":"2021-05-25T15:44:48.752013Z","iopub.status.idle":"2021-05-25T15:44:48.764825Z","shell.execute_reply.started":"2021-05-25T15:44:48.751987Z","shell.execute_reply":"2021-05-25T15:44:48.763875Z"},"trusted":true},"execution_count":17,"outputs":[{"name":"stdout","text":"Model: \"sequential\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nembedding (Embedding)        (None, None, 64)          523840    \n_________________________________________________________________\nbidirectional (Bidirectional (None, None, 128)         66048     \n_________________________________________________________________\nbidirectional_1 (Bidirection (None, None, 64)          41216     \n_________________________________________________________________\nbidirectional_2 (Bidirection (None, 32)                10368     \n_________________________________________________________________\ndense (Dense)                (None, 64)                2112      \n_________________________________________________________________\ndropout (Dropout)            (None, 64)                0         \n_________________________________________________________________\ndense_1 (Dense)              (None, 1)                 65        \n=================================================================\nTotal params: 643,649\nTrainable params: 643,649\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"keras.utils.plot_model(model, show_shapes=True, dpi=48)","metadata":{"execution":{"iopub.status.busy":"2021-05-25T22:55:30.000172Z","iopub.execute_input":"2021-05-25T22:55:30.000469Z","iopub.status.idle":"2021-05-25T22:55:30.172989Z","shell.execute_reply.started":"2021-05-25T22:55:30.000439Z","shell.execute_reply":"2021-05-25T22:55:30.172060Z"},"trusted":true},"execution_count":21,"outputs":[{"execution_count":21,"output_type":"execute_result","data":{"image/png":"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\n","text/plain":"<IPython.core.display.Image object>"},"metadata":{}}]},{"cell_type":"markdown","source":"# Train the Model","metadata":{}},{"cell_type":"code","source":"model.compile(loss='binary_crossentropy',\n             optimizer='adam',\n             metrics=['accuracy'])","metadata":{"execution":{"iopub.status.busy":"2021-05-25T15:44:49.640494Z","iopub.execute_input":"2021-05-25T15:44:49.640767Z","iopub.status.idle":"2021-05-25T15:44:49.662547Z","shell.execute_reply.started":"2021-05-25T15:44:49.640736Z","shell.execute_reply":"2021-05-25T15:44:49.661715Z"},"trusted":true},"execution_count":19,"outputs":[]},{"cell_type":"code","source":"history = model.fit(train_data, \n                    validation_data=test_data, \n                    epochs=2)","metadata":{"execution":{"iopub.status.busy":"2021-05-25T15:44:49.665267Z","iopub.execute_input":"2021-05-25T15:44:49.665592Z","iopub.status.idle":"2021-05-25T22:55:29.996352Z","shell.execute_reply.started":"2021-05-25T15:44:49.665564Z","shell.execute_reply":"2021-05-25T22:55:29.995350Z"},"trusted":true},"execution_count":20,"outputs":[{"name":"stdout","text":"Epoch 1/2\n391/391 [==============================] - 12933s 33s/step - loss: 0.8839 - accuracy: 0.5076 - val_loss: 0.6950 - val_accuracy: 0.5000\nEpoch 2/2\n391/391 [==============================] - 12908s 33s/step - loss: 0.6941 - accuracy: 0.5031 - val_loss: 0.6932 - val_accuracy: 0.5000\n","output_type":"stream"}]},{"cell_type":"code","source":"hist = history.history\n\nplt.plot(hist['loss'])\nplt.plot(hist['val_loss'])\nplt.legend(labels=['Training', 'Validation'])\nplt.xlabel('Epochs')\nplt.ylabel('Loss')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-05-25T22:55:30.175010Z","iopub.execute_input":"2021-05-25T22:55:30.175420Z","iopub.status.idle":"2021-05-25T22:55:30.355031Z","shell.execute_reply.started":"2021-05-25T22:55:30.175345Z","shell.execute_reply":"2021-05-25T22:55:30.354032Z"},"trusted":true},"execution_count":22,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# Sample prediction\n\nsamples = ['The plot is fantastic', \n           'The movie was cool and thrilling', \n           'one of the worst films I have ever seen']\n\n# encode into integers\nsample_encoded = [encoder.encode(sample) for sample in samples]\n\n# pad with zeros to have same length \nsample_padded = []\nfor s in sample_encoded:\n    pad_length = 128 - len(s)\n    zeros = [0]*pad_length\n    s.extend(zeros)\n    s = tf.convert_to_tensor(s)\n    sample_padded.append(s)\n    \n# convert into tensor before feeding the model\nsample_padded = tf.convert_to_tensor(sample_padded)\n#make predictions\npredictions = model.predict(sample_padded)\npredictions","metadata":{"execution":{"iopub.status.busy":"2021-05-26T00:57:18.330189Z","iopub.execute_input":"2021-05-26T00:57:18.330595Z","iopub.status.idle":"2021-05-26T00:57:18.337429Z","shell.execute_reply.started":"2021-05-26T00:57:18.330562Z","shell.execute_reply":"2021-05-26T00:57:18.336191Z"},"trusted":true},"execution_count":23,"outputs":[{"execution_count":23,"output_type":"execute_result","data":{"text/plain":"array([[0.90058655],\n       [0.791466  ],\n       [0.2801157 ]], dtype=float32)"},"metadata":{}}]},{"cell_type":"code","source":"print('Predictions on sample test reviews... \\n')\nfor i in range(len(samples)):\n    pred = predictions[i][0]\n    sentiment = 'positive' if pred>0.5 else 'negative'\n    print('%40s : %s'%(samples[i], sentiment))","metadata":{"execution":{"iopub.status.busy":"2021-05-26T01:11:08.274262Z","iopub.execute_input":"2021-05-26T01:11:08.274698Z","iopub.status.idle":"2021-05-26T01:11:08.281455Z","shell.execute_reply.started":"2021-05-26T01:11:08.274662Z","shell.execute_reply":"2021-05-26T01:11:08.280519Z"},"trusted":true},"execution_count":29,"outputs":[{"name":"stdout","text":"Predictions on sample test reviews... \n\n                   The plot is fantastic : positive\n        The movie was cool and thrilling : positive\n one of the worst films I have ever seen : negative\n","output_type":"stream"}]},{"cell_type":"markdown","source":"https://github.com/RajkumarGalaxy/NLP/blob/master/get-started-with-sentiment-analysis-tf-keras.ipynb\n","metadata":{}},{"cell_type":"markdown","source":"","metadata":{}},{"cell_type":"markdown","source":"### Thank You For Your Time!","metadata":{}}]}