Text Classification
Build an end-to-end pipeline to classify text using ML models on the IMDB dataset.
Run in Google Colab ↗Overview
Text classification is the task of assigning predefined categories to text. In this module, you build an end-to-end pipeline on the IMDB dataset: preprocess reviews, vectorize with TF-IDF, train a classifier, and evaluate performance — connecting every prior module together.
💡 Why It Matters
Classification is the most common applied NLP task — spam detection, sentiment analysis, topic labeling. This module demonstrates how all preprocessing steps compound: better preprocessing → better vectors → better classifier performance.
Pipeline Components
Naive Bayes
A probabilistic classifier that assumes feature independence — fast and strong baseline for text.
Logistic Regression
A linear model that often outperforms Naive Bayes on TF-IDF features with enough data.
scikit-learn Pipeline
Chains vectorizer + classifier into one object for clean training and inference.
Evaluation Metrics
Accuracy, precision, recall, and F1-score — all critical for understanding classifier behavior beyond raw accuracy.
🛠 Library Note
This module uses scikit-learn's `Pipeline`, `TfidfVectorizer`, `MultinomialNB`, and `LogisticRegression` on the IMDB Dataset.csv from this repo.
External Documentation
What You'll Learn
- Building a full preprocessing → vectorization → classification pipeline
- Training Naive Bayes and Logistic Regression on IMDB reviews
- Evaluating with accuracy, precision, recall, and F1