Weather Station API with ML Support Endpoints: POST /update - Full weather data (temp, humidity, pressure, wind) POST /wind/rapid - Rapid wind-only updates (5-second interval) POST /storm/risk - Storm detection data (probability, risk level) POST /ml/inference - ML predictions (storm prob, anomaly, wind forecast) GET /data - Get current weather + ML data GET /data/download - Download current data as JSON file GET /wind - Get wind-specific data GET /history - Get 24h historical data (JSON) GET /history?hours=48 - Get custom hours of history (JSON) GET /history/download - Download 24h data as CSV file GET /history/download?hours=48 - Download custom hours as CSV GET /logs - View system logs (HTML) GET /stats - View storage statistics POST /cleanup - Manual cleanup trigger Delete History: DELETE /history - Delete ALL historical data DELETE /history/range - Delete time range (body: {start_time, end_time}) GET /history/cleanup - Keep last 24 hours (default) GET /history/cleanup?hours=48 - Keep last 48 hours ML Data Fields: ml_storm_probability - Neural network storm prediction (0-100%) ml_anomaly_score - Anomaly detection score (threshold: 0.75) ml_wind_gust_5min - Predicted wind gust 5 minutes ahead ml_wind_gust_10min - Predicted wind gust 10 minutes ahead storm_probability - Blended storm probability storm_risk - Risk level (Low/Moderate/Elevated/High/Severe) pressure_drop_rate - Barometric pressure change (hPa/hour) wind_acceleration - Wind speed change rate (mph/minute) Smart Storage System: FULL UPDATES: 5-min normal, 30-sec events (~288/day) WIND RAPID: 5-sec updates (~17,280/day) ML INFERENCE: 10-sec updates (~8,640/day) STORM DETECTION: 30-sec updates (~2,880/day) Total: ~29,000 operations/day (under 100k limit)