E-Mail Classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Identifying spam e-mails with neural networks\n",
    "\n",
    "A common use for binary classification is sorting spam e-mails from legitimate e-mails. It's not difficult to build a neural network that classifies text. Let's use Keras to build a binary classifier for e-mails, train it with a collection of e-mails labeled with 0s (for not spam) and 1s (for spam), and then run a few e-mails through it to see how well it performs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load and prepare the data\n",
    "\n",
    "The first step is to load the dataset and prepare it for use in deep learning. We'll use Keras's [`Tokenizer`](https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/text/Tokenizer) class to vectorize the text in the training e-mails, and we'll get some help from the [Natural Language Toolkit](https://www.nltk.org/) for removing stop words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>IsSpam</th>\n",
       "      <th>Text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>key issues going forwarda year end reviews rep...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>congrats contratulations the execution the cen...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>key issues going forwardall under control set...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>epmi files protest entergy transcoattached our...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>california power please contact kristin walsh ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   IsSpam                                               Text\n",
       "0       0  key issues going forwarda year end reviews rep...\n",
       "1       0  congrats contratulations the execution the cen...\n",
       "2       0   key issues going forwardall under control set...\n",
       "3       0  epmi files protest entergy transcoattached our...\n",
       "4       0  california power please contact kristin walsh ..."
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('Data/ham-spam.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Find out how many rows the dataset contains and confirm that there are no missing values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 1000 entries, 0 to 999\n",
      "Data columns (total 2 columns):\n",
      " #   Column  Non-Null Count  Dtype \n",
      "---  ------  --------------  ----- \n",
      " 0   IsSpam  1000 non-null   int64 \n",
      " 1   Text    1000 non-null   object\n",
      "dtypes: int64(1), object(1)\n",
      "memory usage: 15.8+ KB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove duplicate rows from the dataset and check for balance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">Text</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>unique</th>\n",
       "      <th>top</th>\n",
       "      <th>freq</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IsSpam</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>499</td>\n",
       "      <td>499</td>\n",
       "      <td>summary points cdwrattached please find some n...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>500</td>\n",
       "      <td>500</td>\n",
       "      <td>hidear sir madam engr munachim eze chairman th...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Text                                                               \n",
       "       count unique                                                top freq\n",
       "IsSpam                                                                     \n",
       "0        499    499  summary points cdwrattached please find some n...    1\n",
       "1        500    500  hidear sir madam engr munachim eze chairman th...    1"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.drop_duplicates()\n",
    "df.groupby('IsSpam').describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a feature column *x* containing the text in the \"Text\" column with stopwords removed, and a label column *y*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.corpus import stopwords\n",
    "from nltk.tokenize import word_tokenize\n",
    "\n",
    "def remove_stop_words(text):\n",
    "    text = word_tokenize(text.lower())\n",
    "    stop_words = set(stopwords.words('english'))\n",
    "    text = [word for word in text if word.isalpha() and not word in stop_words]\n",
    "    return ' '.join(text)\n",
    "    \n",
    "x = df.apply(lambda row: remove_stop_words(row['Text']), axis=1)\n",
    "y = df['IsSpam']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Tokenize the text and create padded sequences from it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "\n",
    "max_words = 20000\n",
    "max_length = 500\n",
    "\n",
    "tokenizer = Tokenizer(num_words=max_words)\n",
    "tokenizer.fit_on_texts(x)\n",
    "sequences = tokenizer.texts_to_sequences(x)\n",
    "x = pad_sequences(sequences, maxlen=max_length)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train a neural network to identify spam\n",
    "\n",
    "Create a neural network containing an [`Embedding`](https://keras.io/api/layers/core_layers/embedding/) layer for converting sequences into arrays of word vectors and a [`Dense`](https://keras.io/api/layers/core_layers/dense/) layer for classifying arrays of word vectors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding (Embedding)        (None, 500, 32)           640000    \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 16000)             0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 128)               2048128   \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1)                 129       \n",
      "=================================================================\n",
      "Total params: 2,688,257\n",
      "Trainable params: 2,688,257\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Flatten\n",
    "from keras.layers.embeddings import Embedding\n",
    "\n",
    "model = Sequential() \n",
    "model.add(Embedding(max_words, 32, input_length=max_length)) \n",
    "model.add(Flatten())\n",
    "model.add(Dense(128, activation='relu'))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy']) \n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "40/40 [==============================] - 8s 15ms/step - loss: 0.6778 - accuracy: 0.6391 - val_loss: 0.7083 - val_accuracy: 0.5900\n",
      "Epoch 2/5\n",
      "40/40 [==============================] - 0s 8ms/step - loss: 0.4587 - accuracy: 0.8197 - val_loss: 0.4924 - val_accuracy: 0.9050\n",
      "Epoch 3/5\n",
      "40/40 [==============================] - 0s 8ms/step - loss: 0.1467 - accuracy: 0.9916 - val_loss: 0.3130 - val_accuracy: 0.9350\n",
      "Epoch 4/5\n",
      "40/40 [==============================] - 0s 8ms/step - loss: 0.0427 - accuracy: 0.9880 - val_loss: 0.1308 - val_accuracy: 0.9750\n",
      "Epoch 5/5\n",
      "40/40 [==============================] - 0s 8ms/step - loss: 0.0186 - accuracy: 0.9963 - val_loss: 0.1083 - val_accuracy: 0.9700\n"
     ]
    }
   ],
   "source": [
    "hist = model.fit(x, y, validation_split=0.2, epochs=5, batch_size=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Chart the training and validation accuracy for each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "sns.set()\n",
    "\n",
    "acc = hist.history['accuracy']\n",
    "val = hist.history['val_accuracy']\n",
    "epochs = range(1, len(acc) + 1)\n",
    "\n",
    "plt.plot(epochs, acc, '-', label='Training accuracy')\n",
    "plt.plot(epochs, val, ':', label='Validation accuracy')\n",
    "plt.title('Training and Validation Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train a convolutional neural network to identify spam\n",
    "\n",
    "Convolutional neural networks (CNNs) are primarily used to classify images, but they can be helpful for text classification, too. One advantage to a CNN is that it can recognize word patterns and factor them into its predictions rather than treat words independently. Let's train a CNN and see if it can achieve a higher validation accuracy than a conventional neural network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_1 (Embedding)      (None, 500, 32)           640000    \n",
      "_________________________________________________________________\n",
      "conv1d (Conv1D)              (None, 494, 32)           7200      \n",
      "_________________________________________________________________\n",
      "max_pooling1d (MaxPooling1D) (None, 98, 32)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_1 (Conv1D)            (None, 92, 32)            7200      \n",
      "_________________________________________________________________\n",
      "global_max_pooling1d (Global (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 654,433\n",
      "Trainable params: 654,433\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import Conv1D, MaxPooling1D, GlobalMaxPooling1D\n",
    "\n",
    "model = Sequential() \n",
    "model.add(Embedding(max_words, 32, input_length=max_length)) \n",
    "model.add(Conv1D(32, 7, activation='relu'))\n",
    "model.add(MaxPooling1D(5))\n",
    "model.add(Conv1D(32, 7, activation='relu'))\n",
    "model.add(GlobalMaxPooling1D())\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy']) \n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "40/40 [==============================] - 2s 14ms/step - loss: 0.6786 - accuracy: 0.5844 - val_loss: 0.9463 - val_accuracy: 0.0000e+00\n",
      "Epoch 2/5\n",
      "40/40 [==============================] - 0s 9ms/step - loss: 0.5787 - accuracy: 0.6234 - val_loss: 0.8060 - val_accuracy: 0.0950\n",
      "Epoch 3/5\n",
      "40/40 [==============================] - 0s 9ms/step - loss: 0.3418 - accuracy: 0.8284 - val_loss: 0.4021 - val_accuracy: 0.9500\n",
      "Epoch 4/5\n",
      "40/40 [==============================] - 0s 9ms/step - loss: 0.1180 - accuracy: 0.9715 - val_loss: 0.1440 - val_accuracy: 0.9900\n",
      "Epoch 5/5\n",
      "40/40 [==============================] - 0s 9ms/step - loss: 0.0358 - accuracy: 0.9917 - val_loss: 0.0852 - val_accuracy: 0.9900\n"
     ]
    }
   ],
   "source": [
    "hist = model.fit(x, y, validation_split=0.2, epochs=5, batch_size=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Chart the training and validation accuracy for each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = hist.history['accuracy']\n",
    "val = hist.history['val_accuracy']\n",
    "epochs = range(1, len(acc) + 1)\n",
    "\n",
    "plt.plot(epochs, acc, '-', label='Training accuracy')\n",
    "plt.plot(epochs, val, ':', label='Validation accuracy')\n",
    "plt.title('Training and Validation Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use the model to classify e-mails\n",
    "\n",
    "Now let's see how the model classifies some sample e-mails. We'll start with a message that is not spam. The model's `predict` method predicts the probability that the input belongs to the positive class (spam)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.32004166"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cleaned_text = remove_stop_words('Can you attend a code review on Tuesday? Need to make sure the logic is rock solid.')\n",
    "sequence = tokenizer.texts_to_sequences([cleaned_text])\n",
    "padded_sequence = pad_sequences(sequence, maxlen=max_length)\n",
    "model.predict(padded_sequence)[0][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now test the model with a spam message."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7966199"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cleaned_text = remove_stop_words('Why pay more for expensive meds when you can order them online and save $$$?')\n",
    "sequence = tokenizer.texts_to_sequences([cleaned_text])\n",
    "padded_sequence = pad_sequences(sequence, maxlen=max_length)\n",
    "model.predict(padded_sequence)[0][0]"
   ]
  }
 ],
 "metadata": {
  "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}