Cross-domain sentiment analysis pipeline comparing classical machine learning (TF-IDF + Logistic Regression) with modern transformer architectures (DistilBERT) to measure out-of-domain performance drops.

End-to-end Twitter sentiment analysis using NLP and Logistic Regression on the Sentiment140 dataset, including text preprocessing, TF-IDF vectorization, model training, evaluation, and persistence.

Fine-tune BERT (bert-base-uncased) for sentiment classification on the Sentiment140 Twitter dataset using the Hugging Face Transformers library with PyTorch.