Spam Detection - Text Classification

Objective

The goal of this project is to classify text data from three different languages (English, French, and German) as either "Spam" or "Ham" (not spam). Using a machine learning model, we aim to accurately classify the dataset based on text features. The dataset is provided in a CSV file named Task1.csv.

Dataset

The dataset consists of text messages labeled as "Spam" or "Ham" in three languages: English, French, and German.

Workflow

1. Data Loading & Preprocessing

• Load the dataset: Use Pandas to load the CSV file into a DataFrame.
• Missing values: Check for missing values and clean or remove them as necessary.
• Balancing the dataset: Ensure the dataset is balanced. If the dataset is imbalanced, apply techniques such as oversampling, undersampling, or using class weights in the model.

2. Feature Extraction

• Vectorization: Use TfidfVectorizer to convert the text data into numerical representations. The vectorizer should handle multiple languages.
• Language-agnostic vectorization: Ensure that the vectorization process works well across all languages (English, French, and German).

3. Model Training

• Multinomial Naïve Bayes: Train a Multinomial Naïve Bayes model using the preprocessed text data. Understand the principles of MultinomialNB to better implement and tune the model.

4. Model Evaluation

• Accuracy: Print the accuracy of the trained model.
• Confusion Matrix: Create a confusion matrix to visualize the model’s classification performance for "Spam" and "Ham" labels.