Energy use anomaly: smart meter spike — leak или just hot day?

Ситуация: Smart meters detect usage spike. Could be: leak, malfunction, или just unusual usage (party, hot day). False positives annoy customers. Design framework.

IoT energy meters. Send readings hourly. Anomaly = sudden spike vs baseline.

Доступные данные

• meter_readings: meter_id, ts, usage_kwh
• weather: region, ts, temperature
• alerts: meter_id, ts, type (anomaly), resolved (false_pos / true_pos)

Задачи

• Anomaly detection: simple threshold vs ML.
• Context: weather, time-of-day, day-of-week.
• False positive reduction.
• Action: alert customer? Auto-investigate?

Все кейсы для подготовки → ← Все кейсы