Sampling noise

What this is

Two single-label datasets — customer-support tickets routed to one of five categories (multi-class), and movie-review sentiment as POSITIVE/NEGATIVE (binary). For each, the page guides you through the tabs: Explore the data (no classifier yet), then the ML classifier (TF-IDF + Logistic Regression), plus a companion lab on noise across multiple metrics. Predictions and ground truth are precomputed, so you can spin freely — no API calls. (The gen-AI classifier that used to live here has moved to the L4 lab, LLM as a Judge.)