Predictive outlook
5-day heatwave forecast
Probability and severity for the next five days, the factors behind them and the observations the forecast reads.
Trajectory
Heatwave probability, next 5 days
Hover a day for severity and expected duration
No forecast available
Latest stored prediction
Ensemble record
The most recent prediction saved for this ward
No stored prediction yet
A prediction is saved each time the 5-day forecast is generated.
Predictive outlook
5-day heatwave forecast
Probability, severity and expected duration per day
Produced by the documented fallback model, which assigns the same probability to each of the five days.
Transparency
Explainable drivers · model statusExplainability
What pushes the probability
Signed share of each factor relative to a mild baseline day
Baselines: 28 °C mean, 32 °C maximum, 45% humidity, 3.5 m/s wind, 350 W/m² solar. Health vulnerability is scaled from the ward's elderly share and impervious surface.
Model status
How this forecast is produced
Fallback model active
The XGBoost, Random Forest and LSTM artifacts are produced by a separate Python training job and are not loaded here. The original code therefore uses its logistic fallback on the 3-day temperature anomaly.
Forward weather leg: not connected
Without an OpenWeather key the forecast reads observed history only.
Validation
The training script computes a test accuracy for each tree model, but no evaluation results are recorded in the project, so none are shown.
Observed history
Last 30 observationsInput series
Mean air temperature
Last 0 daily observations
Input series
Relative humidity
Last 0 daily observations