Model Performance Comparison
About Model Comparison
Compare Gradient Boosting model accuracy by forecast horizon (short-term 0–48h, medium-term 0–7d, long-term 0–28d), accuracy by lead-time, and ensemble vs deterministic approaches. Metrics shown are Mean Absolute Error (MAE) — lower is better.
Short-Term Accuracy (0–48h)
Medium-Term Accuracy (0–7d)
Forecast Horizon Comparison
Ensemble vs Deterministic at Same Lead Times
Fair comparison: Comparing forecasts at 6h, 12h, 24h, and 48h lead timesForecast Accuracy Trends
Performance Summary
| Model / Type | Temperature MAE (°C) | Humidity MAE (%) | Pressure MAE (hPa) | Wind Speed MAE (mph) | Overall Score |
|---|---|---|---|---|---|
| GradientBoosting-LongTerm | 1.09 | 4.81 | 2.12 | 0.38 | 0.19 |
| GradientBoosting-MediumTerm | 2.86 | 7.13 | 3.58 | 0.37 | 0.36 |
| GradientBoosting-ShortTerm | 3.66 | 11.47 | 2.07 | 0.54 | 0.42 |
Long-term models appear here once their forecast times (7–28 days out) have passed and been verified.