Anti-cheat: 5% подозрительных аккаунтов — false-positive problem

Ситуация: Multiplayer game. ML detector flags 5% accounts as cheaters. Legit players banned, refund requests. Tune detection.

Multiplayer MOBA. 1M MAU. Cheating: aimbot, wallhack, account boosting. Ban = appeal possible.

Доступные данные

• accounts: id, registered_at, ban_status
• detections: account_id, ts, signal_type, score
• appeals: account_id, ts, outcome

Задачи

• False positive rate analysis.
• Per-signal type accuracy.
• Recalibration.
• Process improvements.

Все кейсы для подготовки → ← Все кейсы