Identifying spam emails is an important task for individuals and organizations, as it helps them protect their privacy and security. There are a number of characteristics that can be used to identify spam emails, such as the presence of certain words or phrases, the sender’s email address, or the format of the email.
In this blog post, we will develop an advanced machine learning model to identify spam emails using Python. We will start by collecting and preprocessing the data, and then we will train and evaluate a model using a number of different techniques.
Data collection and preprocessing:
To create our machine learning model, we will need a dataset of emails that we can use to train the model. There are a number of publicly available datasets that we can use for this purpose, such as the “Spam Email Classification” dataset from Kaggle.
To begin, we will import the necessary libraries and download the dataset using Pandas.
import pandas as pd url = "https://www.kaggle.com/uciml/sms-spam-collection-dataset/download" data = pd.read_csv(url, sep='\t', header=None, names=["label", "text"])
Next, we will split the data into features and the target variable. The features will be the text of the email, and the target variable will be whether or not the email is spam.
X = data["text"] y = data["label"]
To prepare the data for training, we will need to preprocess the text of the emails. This can be done using techniques such as stemming, lemmatization, and stop word removal.
We can perform these techniques using the nltk
library.
import nltk from nltk.stem import PorterStemmer from nltk.corpus import stopwords from nltk.tokenize import word_tokenize # Initialize stemmer and stop words stemmer = PorterStemmer() stop_words = set(stopwords.words("english")) # Preprocess text X_processed = [] for text in X: # Remove punctuation and lowercase text = text.translate(str.maketrans("", "", string.punctuation)).lower() # Tokenize and stem words tokens = [stemmer.stem(word) for word in word_tokenize(text) if word not in stop_words] # Rejoin stemmed words text = " ".join(tokens) X_processed.append(text)
Model development:
Now that our data is prepared, we can start developing our machine learning model. There are a number of different algorithms that we could use for this task, such as naive Bayes or support vector machines (SVMs). For this example, we will use a deep learning model, specifically a long short-term memory (LSTM) neural network.
LSTM networks are particularly well-suited for natural language processing tasks, as they are able to capture long-term dependencies in the data. To create the LSTM model, we will use the `Sequential` and `LSTM` classes from the `keras` library.
```python from keras.models import Sequential from keras.layers import Embedding, LSTM, Dense # Create LSTM model model = Sequential() model.add(Embedding(input_dim=vocab_size, output_dim=128)) model.add(LSTM(units=128, dropout=0.5, recurrent_dropout=0.5)) model.add(Dense(units=1, activation="sigmoid"))
Model evaluation:
Now that our model is created, we can train it on the data using the fit
method.
# Train model model.fit(X_train, y_train, batch_size=32, epochs=10)
After the model is trained, we can evaluate its performance on the test data. We will start by making predictions on the test data using the predict
method of the Sequential
object.
# Predict labels for test data y_pred = model.predict(X_test)
To evaluate the model’s performance, we can use a number of different metrics. One common metric is the accuracy, which measures the proportion of correct predictions made by the model.
We can calculate the accuracy using the accuracy_score
function from Scikit-learn.
from sklearn.metrics import accuracy_score accuracy = accuracy_score(y_test, y_pred) print("Accuracy:", accuracy)
Another metric we can use is the confusion matrix, which shows the number of true positive, true negative, false positive, and false negative predictions made by the model.
We can create the confusion matrix using the confusion_matrix
function from Scikit-learn.
from sklearn.metrics import confusion_matrix confusion_matrix = confusion_matrix(y_test, y_pred) print("Confusion matrix:", confusion_matrix)
To better understand the model’s performance, we can also create a classification report, which includes precision, recall, and f1-score metrics.
We can create the classification report using the classification_report
function from Scikit-learn.
from sklearn.metrics import classification_report report = classification_report(y_test, y_pred) print("Classification report:", report) from sklearn.metrics import classification_report report = classification_report(y_test, y_pred) print("Classification report:", report)
Here is the complete code for creating an advanced machine learning model to identify spam emails using Python:
import pandas as pd import nltk import string from nltk.stem import PorterStemmer from nltk.corpus import stopwords from nltk.tokenize import word_tokenize from sklearn.model_selection import train_test_split from keras.preprocessing.text import Tokenizer from keras.preprocessing.sequence import pad_sequences from keras.models import Sequential from keras.layers import Embedding, LSTM, Dense from sklearn.metrics import accuracy_score, confusion_matrix, classification_report # Download and load dataset url = "https://www.kaggle.com/uciml/sms-spam-collection-dataset/download" data = pd.read_csv(url, sep='\t', header=None, names=["label", "text"]) # Split data into features and target variable X = data["text"] y = data["label"] # Initialize stemmer and stop words stemmer = PorterStemmer() stop_words = set(stopwords.words("english")) # Preprocess text X_processed = [] for text in X: # Remove punctuation and lowercase text = text.translate(str.maketrans("", "", string.punctuation)).lower() # Tokenize and stem words tokens = [stemmer.stem(word) for word in word_tokenize(text) if word not in stop_words] # Rejoin stemmed words text = " ".join(tokens) X_processed.append(text) # Tokenize and pad sequences tokenizer = Tokenizer() X_tokenized = tokenizer.fit_transform(X_processed) X_padded = pad_sequences(X_tokenized, maxlen=500) # Split data into train and test sets X_train, X_test, y_train, y_test = train_test_split(X_padded, y, test_size=0.2) # Create LSTM model model = Sequential() model.add(Embedding(input_dim=vocab_size, output_dim=128)) model.add(LSTM(units=128, dropout=0.5, recurrent_dropout=0.5)) model.add(Dense(units=1, activation="sigmoid")) # Train model model.fit(X_train, y_train, batch_size=32, epochs=10) #Predict labels for test data y_pred = model.predict(X_test) #Calculate accuracy accuracy = accuracy_score(y_test, y_pred) print("Accuracy:", accuracy) #Create confusion matrix confusion_matrix = confusion_matrix(y_test, y_pred) print("Confusion matrix:", confusion_matrix) #Create classification report report = classification_report(y_test, y_pred) print("Classification report:", report)