Sentiment analysis is a classification task where text is categorized (e.g., positive, negative, or neutral).
Logistic regression is a fundamental and highly effective statistical method for binary or multi-class classification, making it a standard choice for sentiment models. It estimates the probability of an input belonging to a specific class. While more complex architectures like Transformers (LLMs) are now common, logistic regression remains a benchmark for its simplicity and transparency in audit contexts. Image recognition and ANOVA are not relevant to the linguistic processing required for sentiment analysis, and the log-rank test is typically used in survival analysis rather than NLP classification.