How does Naïve Bayes handle online learning and streaming updates mathematically?

Naïve Bayes updates sufficient statistics incrementally, making it ideal for streaming data.

In online learning, NB updates class counts and feature counts incrementally: new_count = old_count + observed_count. For Gaussian NB, means and variances are updated using recursive formulas. This avoids retraining and supports real-time adaptation in evolving environments.