A sentiment analysis pipeline for classifying hotel reviews as positive or negative, using models ranging from traditional ML to BERT-based Transformers.
- 📊 Multi-model benchmark: Logistic Regression, CNN, BiLSTM, BERT, LSTM, and Tranformers
- 🧼 Full preprocessing pipeline: tokenization, stopword removal, lemmatization
- 🤗 Integrated with HuggingFace Transformers
- 🔁 Custom training loop with gradient clipping and DataLoader
- 📈 Visualizations: word cloud, confusion matrix, accuracy plots