get-next-word.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Science - Micro Project\n",
    "**Course**: CS1U31G (_Applied Data Science with Python_)\n",
    "\n",
    "**Submitted By**: Abhinav Prakash (B21CS1203)\n",
    "\n",
    "**Topic**: Using a deep learning based sequence model to predict the next likely words given an incomplete sentence as the input.\n",
    "\n",
    "## Overview\n",
    "\n",
    "* ### [1. Load the Data](#load-the-data)\n",
    "* ### [2. Overview of the Data](#data-cleaning)\n",
    "* ### [3. Exploratory Data Analysis](#exploratory-data-analysis)\n",
    "* ### [4. Data Preparation](#data-preparation)\n",
    "* ### [5. Model Building](#model-building)\n",
    "* ### [6. Improve Model](#improve-model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. Load the Data <a class=\"anchor\" id=\"load-the-data\"></a>\n",
    "\n",
    "We use the popular drama/indie screenplay: *Good Will Hunting* as our training corpus. This corpus comprises of 3293 unique words in total."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cIBfSFOs5dft"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
    "from tensorflow.keras.layers import Embedding, LSTM, Dense, Bidirectional, Flatten\n",
    "from tensorflow.keras.preprocessing.text import Tokenizer\n",
    "from tensorflow.keras.models import Sequential\n",
    "\n",
    "import numpy as np "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nbXB5xdMQUqp"
   },
   "outputs": [],
   "source": [
    "file = open(\"./corpus/Goodwill.txt\").read()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Overview of the Data <a class=\"anchor\" id=\"overview-of-the-data\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.1 Descriptive Statistics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "KWFqIoL38c9M"
   },
   "outputs": [],
   "source": [
    "data = file.lower().split(\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "4CmX4_qrAxXz",
    "outputId": "7c5b36b7-2022-4ac2-d511-d4d5e1ab78fd"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['   good will hunting',\n",
       " '           by',\n",
       " 'matt damon & ben affleck',\n",
       " '',\n",
       " 'fade in:',\n",
       " \"ext. south boston st. patrick's day parade -- day cut to:\",\n",
       " 'int. l street bar & grille, south boston -- evening',\n",
       " \"the bar is dirty, more than a little run down. if there is ever a cook on duty, he's not here now. as we pan across several empty tables, we can almost smell the odor of last nights beer and crushed pretzels on the floor.\",\n",
       " '                               chuckie',\n",
       " '            oh my god, i got the most fucked up',\n",
       " \"thing i been meanin' to tell you.\",\n",
       " 'as the camera rises, we find four young men seated around a table near the back of the bar.',\n",
       " 'all oh jesus. here we go.',\n",
       " \"the guy holding court is chuckie sullivan, 20, and the largest of the bunch. he is loud, boisterous, a born entertainer. next to him is will hunting, 20, handsome and confident, a soft- spoken leader. on will's right sits billy mcbride, 22, heavy, quiet, someone you definitely wouldn't want to tangle with. finally there is morgan o'mally, 19, smaller than the other\",\n",
       " \"guys. wiry and anxious, morgan listens to chuckie's stories with eager disgust.\",\n",
       " 'all four boys speak with thick boston accents. this working class irish neighborhood and these boys are',\n",
       " '                               chuckie',\n",
       " '            you guys know my cousin mikey sullivan?',\n",
       " 'all',\n",
       " 'yeah.']"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No of lines: 3637\n"
     ]
    }
   ],
   "source": [
    "print(f\"No of lines: {len(data)}\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Exploratory Data Analysis <a class=\"anchor\" id=\"exploratory-data-analysis\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "from collections import Counter\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_histogram(words_freq):\n",
    "    words, freq = zip(*words_freq)\n",
    "    plt.figure(figsize=(10, 6))\n",
    "    plt.bar(words, freq)\n",
    "    plt.xlabel('Words')\n",
    "    plt.ylabel('Frequency')\n",
    "    plt.title('Word Frequency Histogram')\n",
    "    plt.xticks(rotation=45, ha='right')\n",
    "    plt.tight_layout()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[('the', 856), ('will', 765), ('you', 763), ('i', 684), ('a', 625), ('to', 619), ('and', 489), ('s', 488), ('sean', 367), ('t', 296), ('that', 295), ('it', 283), ('lambeau', 279), ('of', 276), ('is', 273), ('in', 267), ('chuckie', 235), ('he', 234), ('skylar', 175), ('d', 168)]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "words = []\n",
    "for line in data:\n",
    "  words.extend(re.findall(r'\\b\\w+\\b', line.lower()))\n",
    "\n",
    "word_freq = Counter(words)\n",
    "sorted_word_freq = sorted(word_freq.items(), key=lambda x: x[1], reverse=True)\n",
    "print(f\"{sorted_word_freq[:20]}\")\n",
    "plot_histogram(sorted_word_freq[:20])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4. Data Preparation <a class=\"anchor\" id=\"data-preparation\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1 Data Cleaning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1.1 Removing Whitespace from the data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "43bGa4FxtGMh"
   },
   "outputs": [],
   "source": [
    "corpus = []\n",
    "for line in data:\n",
    "    a = line.strip()\n",
    "    corpus.append(a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "r8hHTCTrtqrs",
    "outputId": "88253b6b-b97c-4834-87a8-4e8a59538c6a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['good will hunting',\n",
       " 'by',\n",
       " 'matt damon & ben affleck',\n",
       " '',\n",
       " 'fade in:',\n",
       " \"ext. south boston st. patrick's day parade -- day cut to:\",\n",
       " 'int. l street bar & grille, south boston -- evening',\n",
       " \"the bar is dirty, more than a little run down. if there is ever a cook on duty, he's not here now. as we pan across several empty tables, we can almost smell the odor of last nights beer and crushed pretzels on the floor.\",\n",
       " 'chuckie',\n",
       " 'oh my god, i got the most fucked up',\n",
       " \"thing i been meanin' to tell you.\",\n",
       " 'as the camera rises, we find four young men seated around a table near the back of the bar.',\n",
       " 'all oh jesus. here we go.',\n",
       " \"the guy holding court is chuckie sullivan, 20, and the largest of the bunch. he is loud, boisterous, a born entertainer. next to him is will hunting, 20, handsome and confident, a soft- spoken leader. on will's right sits billy mcbride, 22, heavy, quiet, someone you definitely wouldn't want to tangle with. finally there is morgan o'mally, 19, smaller than the other\",\n",
       " \"guys. wiry and anxious, morgan listens to chuckie's stories with eager disgust.\",\n",
       " 'all four boys speak with thick boston accents. this working class irish neighborhood and these boys are',\n",
       " 'chuckie',\n",
       " 'you guys know my cousin mikey sullivan?',\n",
       " 'all',\n",
       " 'yeah.']"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corpus[0:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1.2 Tokenizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-6sHviXZ8ZMc"
   },
   "outputs": [],
   "source": [
    "tokenizer = Tokenizer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 71
    },
    "colab_type": "code",
    "id": "f3LTJT9X8l4z",
    "outputId": "bac92d19-312a-4943-f863-0a5d07843090"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[('the', 1), ('will', 2), ('you', 3), ('a', 4), ('to', 5), ('i', 6), ('and', 7), ('sean', 8), ('of', 9), ('is', 10), ('lambeau', 11), ('in', 12), ('that', 13), ('chuckie', 14), ('he', 15), ('it', 16), ('skylar', 17), ('me', 18), ('with', 19), ('on', 20)]\n",
      "3293\n"
     ]
    }
   ],
   "source": [
    "tokenizer.fit_on_texts(corpus)\n",
    "total_words = len(tokenizer.word_index) + 1\n",
    "\n",
    "print(list(tokenizer.word_index.items())[:20])\n",
    "print(total_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1.2 Creating Predictors and Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JUs1W__K8m_p"
   },
   "outputs": [],
   "source": [
    "input_sequences = []\n",
    "for line in corpus:\n",
    "\ttoken_list = tokenizer.texts_to_sequences([line])[0]\n",
    "\tfor i in range(1, len(token_list)):\n",
    "\t\tn_gram_sequence = token_list[:i+1]\n",
    "\t\tinput_sequences.append(n_gram_sequence)\n",
    "\n",
    "# Padding the sequences\n",
    "max_sequence_len = max([len(x) for x in input_sequences])\n",
    "input_sequences = np.array(pad_sequences(input_sequences, maxlen=max_sequence_len, padding='pre'))\n",
    "\n",
    "# Creating predictors and label\n",
    "xs, labels = input_sequences[:,:-1],input_sequences[:,-1]\n",
    "ys = tf.keras.utils.to_categorical(labels, num_classes=total_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5. Model Building <a class=\"anchor\" id=\"model-building\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "3l_5ID8B841w"
   },
   "outputs": [],
   "source": [
    "model = Sequential([Embedding(total_words, 100, input_length = max_sequence_len-1),\n",
    "                    Bidirectional(LSTM(150)),\n",
    "                    Dense(total_words, activation = 'softmax')])\n",
    "\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.1 Model Training "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "ttwyfIVdR1yV",
    "outputId": "014f530b-abec-470b-c48c-a733bd8b0746"
   },
   "outputs": [],
   "source": [
    "history = model.fit(xs, ys, epochs=30, verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.2 Model Prediction\n",
    "\n",
    "Creating a function to predict the next word given a seed sentence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Y6bfoQ6KG9Ls"
   },
   "outputs": [],
   "source": [
    "def next_word(seed):\n",
    "  seed_text = seed\n",
    "  next_words = 1\n",
    "  for _ in range(next_words):\n",
    "    token_list = tokenizer.texts_to_sequences([seed_text])[0]\n",
    "    token_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')\n",
    "    predicted = model.predict_classes(token_list, verbose=0)\n",
    "    output_word = \"\"\n",
    "    for word, index in tokenizer.word_index.items():\n",
    "      if index == predicted:\n",
    "        output_word = word\n",
    "        break\n",
    "    seed_text += \" \" + output_word\n",
    "  print(seed_text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "l-efvI9dITrF",
    "outputId": "8b589ba7-1463-4d21-f507-f5f4fa911143"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Lets go out\n"
     ]
    }
   ],
   "source": [
    "next_word('Lets go')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "S4FDwha5IiS4",
    "outputId": "6c8ecc2b-b65c-4337-895b-a0013888fef8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I want to meet your\n"
     ]
    }
   ],
   "source": [
    "next_word('I want to meet')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "xZO_3gyDIqqT",
    "outputId": "a5f519c6-4a3b-4d3b-ec28-edc7eb60608d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Congratulations thank you\n"
     ]
    }
   ],
   "source": [
    "next_word('Congratulations')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "sjWFbiWCIynw",
    "outputId": "cb971b6b-312d-46ae-d982-b3d303d2519b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Where are  you from\n"
     ]
    }
   ],
   "source": [
    "next_word('Where are ')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.3 Model Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3.1 Plotting the training accuracy of the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "colab_type": "code",
    "id": "g6nSgySJUrcf",
    "outputId": "2473935b-613d-48c4-eca4-d8b343dd615a"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['accuracy'])\n",
    "plt.xlabel(\"Number of Epochs\")\n",
    "plt.ylabel('Training accuracy per epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "colab_type": "code",
    "id": "r-2NBk6Mjd7A",
    "outputId": "2964ca8a-f164-43f0-8de3-a0d8ebb33630"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'])\n",
    "plt.xlabel(\"Number of Epochs\")\n",
    "plt.ylabel('Training loss per epoch')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 6. Improve Model <a class=\"anchor\" id=\"improve-model\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.1 Hyper-Parameter Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras_tuner as kt\n",
    "from keras.optimizers import Adam\n",
    "from keras.losses import SparseCategoricalCrossentropy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model_builder(hp):\n",
    "  model = Sequential()\n",
    "  model.add(Flatten(input_shape=(28, 28)))\n",
    "\n",
    "  # Tune the number of units in the first Dense layer\n",
    "  # Choose an optimal value between 32-512\n",
    "  hp_units = hp.Int('units', min_value=32, max_value=512, step=32)\n",
    "  model.add(Dense(units=hp_units, activation='relu'))\n",
    "  model.add(Dense(10))\n",
    "\n",
    "  # Tune the learning rate for the optimizer\n",
    "  # Choose an optimal value from 0.01, 0.001, or 0.0001\n",
    "  hp_learning_rate = hp.Choice('learning_rate', values=[1e-2, 1e-3, 1e-4])\n",
    "\n",
    "  model.compile(optimizer=Adam(learning_rate=hp_learning_rate),\n",
    "                loss=SparseCategoricalCrossentropy(from_logits=True),\n",
    "                metrics=['accuracy'])\n",
    "\n",
    "  return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuner = kt.Hyperband(model_builder,\n",
    "                     objective='val_accuracy',\n",
    "                     max_epochs=10,\n",
    "                     factor=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuner.search(xs, ys, epochs=30, validation_split=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_hps = tuner.get_best_hyperparameters(num_trials=1)[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we create an updated model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tuner.hypermodel.build(best_hps)\n",
    "history = model.fit(xs, ys, epochs=30, validation_split=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.2 Save the final model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save(\"./model/next-word-model.h5\")"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "name": "Automatic Text Generator.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}